github
/
ColossalAI
mirror of https://github.com/hpcaitech/ColossalAI
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[doc] update Coati README (#4405) 

* style: apply formatter

* fix: add outdated warnings

* docs: add dataset format and polish

* docs: polish README

* fix: fix json format

* fix: fix typos

* revert: revert 7b example
pull/4436/head
Wenhao Chen 2023-08-14 15:26:27 +08:00 committed by GitHub
parent d86ddd9b29
commit 6d41c3f2aa
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
9 changed files with 527 additions and 354 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

126
applications/Chat/README.md
View File

@ -4,7 +4,6 @@
<span>ColossalChat</span>
</h1>
## Table of Contents
- [Table of Contents](#table-of-contents)
@ -34,7 +33,9 @@
- [Authors](#authors)
- [Citations](#citations)
- [Licenses](#licenses)
---
## What is ColossalChat and Coati ?
[ColossalChat](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat) is the project to implement LLM with RLHF, powered by the [Colossal-AI](https://github.com/hpcaitech/ColossalAI) project.
@ -42,6 +43,7 @@
Coati stands for `ColossalAI Talking Intelligence`. It is the name for the module implemented in this project and is also the name of the large language model developed by the ColossalChat project.
The Coati package provides a unified large language model framework that has implemented the following functions
- Supports comprehensive large-model training acceleration capabilities for ColossalAI, without requiring knowledge of complex distributed training algorithms
- Supervised datasets collection
- Supervised instructions fine-tuning
@ -56,17 +58,19 @@ The Coati package provides a unified large language model framework that has imp
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chatgpt/chatgpt.png" width=700/>
</p>
Image source: https://openai.com/blog/chatgpt
Image source: https://openai.com/blog/chatgpt
</div>
**As Colossal-AI is undergoing some major updates, this project will be actively maintained to stay in line with the Colossal-AI project.**
More details can be found in the latest news.
* [2023/03] [ColossalChat: An Open-Source Solution for Cloning ChatGPT With a Complete RLHF Pipeline](https://medium.com/@yangyou_berkeley/colossalchat-an-open-source-solution-for-cloning-chatgpt-with-a-complete-rlhf-pipeline-5edf08fb538b)
* [2023/02] [Open Source Solution Replicates ChatGPT Training Process! Ready to go with only 1.6GB GPU Memory](https://www.hpc-ai.tech/blog/colossal-ai-chatgpt)
- [2023/03] [ColossalChat: An Open-Source Solution for Cloning ChatGPT With a Complete RLHF Pipeline](https://medium.com/@yangyou_berkeley/colossalchat-an-open-source-solution-for-cloning-chatgpt-with-a-complete-rlhf-pipeline-5edf08fb538b)
- [2023/02] [Open Source Solution Replicates ChatGPT Training Process! Ready to go with only 1.6GB GPU Memory](https://www.hpc-ai.tech/blog/colossal-ai-chatgpt)
## Online demo
<div align="center">
<a href="https://www.youtube.com/watch?v=HcTiHzApHm0">
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/ColossalChat%20YouTube.png" width="700" />
@ -83,13 +87,13 @@ More details can be found in the latest news.
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/ColossalChat%20Speed.jpg" width=450/>
</p>
> DeepSpeedChat performance comes from its blog on 2023 April 12, ColossalChat performance can be reproduced on an AWS p4d.24xlarge node with 8 A100-40G GPUs with the following command: torchrun --standalone --nproc_per_node 8 benchmark_opt_lora_dummy.py --num_collect_steps 1 --use_kernels --strategy colossalai_zero2 --experience_batch_size 64 --train_batch_size 32
> DeepSpeedChat performance comes from its blog on 2023 April 12, ColossalChat performance can be reproduced on an AWS p4d.24xlarge node with 8 A100-40G GPUs with the following command: `torchrun --standalone --nproc_per_node 8 benchmark_opt_lora_dummy.py --num_collect_steps 1 --use_kernels --strategy colossalai_zero2 --experience_batch_size 64 --train_batch_size 32`
## Install
### Install the environment
```shell
```bash
conda create -n coati
conda activate coati
git clone https://github.com/hpcaitech/ColossalAI.git
@ -99,7 +103,7 @@ pip install .
### Install the Transformers
```shell
```bash
pip install transformers==4.30.2
```
@ -107,10 +111,11 @@ pip install transformers==4.30.2
### Supervised datasets collection
we collected 104K bilingual datasets of Chinese and English, and you can find the datasets in this repo
[InstructionWild](https://github.com/XueFuzhao/InstructionWild)
We collected 104K bilingual datasets of Chinese and English, and you can find the datasets in this repo
[InstructionWild](https://github.com/XueFuzhao/InstructionWild) and in this [file](https://github.com/XueFuzhao/InstructionWild/blob/main/data/README.md).
Here is how we collected the data
<p align="center">
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/data-collect.png" width=500/>
</p>
@ -122,6 +127,20 @@ Stage1 is supervised instructs fine-tuning, which uses the datasets mentioned ea
You can run the `examples/train_sft.sh` to start a supervised instructs fine-tuning.
[[Stage1 tutorial video]](https://www.youtube.com/watch?v=-qFBZFmOJfg)
**Note**: the supervised dataset follows the following format,
```json
[
{
"instruction": "Provide a list of the top 10 most popular mobile games in Asia",
"input": "",
"output": "The top 10 most popular mobile games in Asia are:\n1) PUBG Mobile\n2) Pokemon Go\n3) Candy Crush Saga\n4) Free Fire\n5) Clash of Clans\n6) Mario Kart Tour\n7) Arena of Valor\n8) Fantasy Westward Journey\n9) Subway Surfers\n10) ARK Survival Evolved",
"id": 0
},
...
]
```
### RLHF Training Stage2 - Training reward model
Stage2 trains a reward model, which obtains corresponding scores by manually ranking different outputs for the same prompt and supervises the training of the reward model
@ -140,13 +159,46 @@ Stage3 uses reinforcement learning algorithm, which is the most complex part of
You can run the `examples/train_prompts.sh` to start training PPO with human feedback.
[[Stage3 tutorial video]](https://www.youtube.com/watch?v=Z8wwSHxPL9g)
**Note**: the required datasets follow the following format,
- `pretrain dataset`
```json
[
{
"instruction": "Provide a list of the top 10 most popular mobile games in Asia",
"input": "",
"output": "The top 10 most popular mobile games in Asia are:\n1) PUBG Mobile\n2) Pokemon Go\n3) Candy Crush Saga\n4) Free Fire\n5) Clash of Clans\n6) Mario Kart Tour\n7) Arena of Valor\n8) Fantasy Westward Journey\n9) Subway Surfers\n10) ARK Survival Evolved",
"id": 0
},
...
]
```
- `prompt dataset`
```json
[
{
"instruction": "Edit this paragraph to make it more concise: \"Yesterday, I went to the store and bought some things. Then, I came home and put them away. After that, I went for a walk and met some friends.\"",
"id": 0
},
{
"instruction": "Write a descriptive paragraph about a memorable vacation you went on",
"id": 1
},
...
]
```
For more details, see [`examples/`](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples).
### Inference Quantization and Serving - After Training
We provide an online inference server and a benchmark. We aim to run inference on single GPU, so quantization is essential when using large models.
We support 8-bit quantization (RTN), 4-bit quantization (GPTQ), and FP16 inference. You can
We support 8-bit quantization (RTN), 4-bit quantization (GPTQ), and FP16 inference.
Online inference server scripts can help you deploy your own services.
For more details, see [`inference/`](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/inference).
@ -158,6 +210,7 @@ For more details, see [`inference/`](https://github.com/hpcaitech/ColossalAI/tre
<details><summary><b>E-mail</b></summary>
![phd](https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/Phd.png)
</details>
<details><summary><b>coding</b></summary>
@ -191,6 +244,7 @@ For more details, see [`inference/`](https://github.com/hpcaitech/ColossalAI/tre
</details>
### Open QA
<details><summary><b>Game</b></summary>
![Game](https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/game.png)
@ -224,6 +278,7 @@ For more details, see [`inference/`](https://github.com/hpcaitech/ColossalAI/tre
You can find more examples in this [repo](https://github.com/XueFuzhao/InstructionWild/blob/main/comparison.md).
### Limitation
<details><summary><b>Limitation for LLaMA-finetuned models</b></summary>
- Both Alpaca and ColossalChat are based on LLaMA. It is hard to compensate for the missing knowledge in the pre-training stage.
- Lack of counting ability: Cannot count the number of items in a list.
@ -247,7 +302,7 @@ You can find more examples in this [repo](https://github.com/XueFuzhao/Instructi
We have integrated the Transformers save and load pipeline, allowing users to freely call Hugging Face's language models and save them in the HF format.
```
```python
from coati.models.llama import LlamaLM
from coati.trainer import SFTTrainer
@ -256,20 +311,20 @@ tokenizer = AutoTokenizer.from_pretrained(args.pretrain)
(model, optim) = strategy.prepare((model, optim))
trainer = SFTTrainer(model=model,
strategy=strategy,
optim=optim,
train_dataloader=train_dataloader,
eval_dataloader=eval_dataloader,
batch_size=args.batch_size,
max_epochs=args.max_epochs,
accumulation_steps = args.accumulation_steps
)
strategy=strategy,
optim=optim,
train_dataloader=train_dataloader,
eval_dataloader=eval_dataloader,
batch_size=args.batch_size,
max_epochs=args.max_epochs,
accumulation_steps=args.accumulation_steps
)
trainer.fit()
# this saves in pytorch format
strategy.save_model(model, args.save_path, only_rank0=True)
# this saves in HF format. ColossalAI strategy with stage-3 doesn't support this method
# this saves in HF format
strategy.save_pretrained(model, args.save_path, only_rank0=True, tokenizer=tokenizer)
```
@ -280,12 +335,13 @@ strategy.save_pretrained(model, args.save_path, only_rank0=True, tokenizer=token
Here are some examples that can allow you to train a 7B model on a single or multiple consumer-grade GPUs.
If you only have a single 24G GPU, you can use the following script. `batch_size`, `lora_rank` and `grad_checkpoint` are the most important parameters to successfully train the model.
```
```bash
// [INFO]: MAX GPU MEMORY ALLOCATED: 19148.9345703125 MB
torchrun --standalone --nproc_per_node=1 train_sft.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy ddp \
--log_interval 10 \
--save_path /path/to/Coati-7B \
--dataset /path/to/data.json \
--batch_size 1 \
@ -298,12 +354,12 @@ torchrun --standalone --nproc_per_node=1 train_sft.py \
```
`colossalai_gemini` strategy can enable a single 24G GPU to train the whole model without using LoRA if you have sufficient CPU memory. You can use the following script.
```
```bash
torchrun --standalone --nproc_per_node=1 train_sft.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy colossalai_gemini \
--log_interval 10 \
--save_path /path/to/Coati-7B \
--dataset /path/to/data.json \
--batch_size 1 \
@ -315,12 +371,12 @@ torchrun --standalone --nproc_per_node=1 train_sft.py \
```
If you have 4x32 GB GPUs, you can even train the whole 7B model using our `colossalai_zero2_cpu` strategy! The script is given as follows.
```
```bash
torchrun --standalone --nproc_per_node=4 train_sft.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy colossalai_zero2_cpu \
--log_interval 10 \
--save_path /path/to/Coati-7B \
--dataset /path/to/data.json \
--batch_size 1 \
@ -330,8 +386,8 @@ torchrun --standalone --nproc_per_node=4 train_sft.py \
--max_epochs 1 \
--grad_checkpoint
```
</details>
</details>
## The Plan
@ -346,24 +402,26 @@ torchrun --standalone --nproc_per_node=4 train_sft.py \
- [ ] support chain-of-thought by [langchain](https://github.com/hwchase17/langchain)
### Real-time progress
You will find our progress in github project broad
[Coati](https://github.com/orgs/hpcaitech/projects/17/views/1)
You will find our progress in github [project broad](https://github.com/orgs/hpcaitech/projects/17/views/1).
## Invitation to open-source contribution
Referring to the successful attempts of [BLOOM](https://bigscience.huggingface.co/) and [Stable Diffusion](https://en.wikipedia.org/wiki/Stable_Diffusion), any and all developers and partners with computing powers, datasets, models are welcome to join and build the Colossal-AI community, making efforts towards the era of big AI models from the starting point of replicating ChatGPT!
You may contact us or participate in the following ways:
1. [Leaving a Star ⭐](https://github.com/hpcaitech/ColossalAI/stargazers) to show your like and support. Thanks!
2. Posting an [issue](https://github.com/hpcaitech/ColossalAI/issues/new/choose), or submitting a PR on GitHub follow the guideline in [Contributing](https://github.com/hpcaitech/ColossalAI/blob/main/CONTRIBUTING.md).
3. Join the Colossal-AI community on
[Slack](https://join.slack.com/t/colossalaiworkspace/shared_invite/zt-z7b26eeb-CBp7jouvu~r0~lcFzX832w),
and [WeChat(微信)](https://raw.githubusercontent.com/hpcaitech/public_assets/main/colossalai/img/WeChat.png "qrcode") to share your ideas.
[Slack](https://join.slack.com/t/colossalaiworkspace/shared_invite/zt-z7b26eeb-CBp7jouvu~r0~lcFzX832w),
and [WeChat(微信)](https://raw.githubusercontent.com/hpcaitech/public_assets/main/colossalai/img/WeChat.png "qrcode") to share your ideas.
4. Send your official proposal to email contact@hpcaitech.com
Thanks so much to all of our amazing contributors!
## Quick Preview
<div align="center">
<a href="https://chat.colossalai.org/">
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/colossalai/img/Chat-demo.png" width="700" />
@ -397,18 +455,22 @@ Thanks so much to all of our amazing contributors!
| Better Cases | 38 ⚔ **41** | **45** ⚔ 33 |
| Win Rate | 48% ⚔ **52%** | **58%** ⚔ 42% |
| Average Score | 7.06 ⚔ **7.13** | **7.31** ⚔ 6.82 |
- Our Coati-7B model performs better than Alpaca-7B when using GPT-4 to evaluate model performance. The Coati-7B model we evaluate is an old version we trained a few weeks ago and the new version is around the corner.
## Authors
Coati is developed by ColossalAI Team:
- [Fazzie](https://fazzie-key.cool/about/index.html)
- [FrankLeeeee](https://github.com/FrankLeeeee)
- [BlueRum](https://github.com/ht-zhou)
- [ver217](https://github.com/ver217)
- [ofey404](https://github.com/ofey404)
- [Wenhao Chen](https://github.com/CWHer)
The Phd student from [(HPC-AI) Lab](https://ai.comp.nus.edu.sg/) also contributed a lot to this project.
- [Zangwei Zheng](https://github.com/zhengzangw)
- [Xue Fuzhao](https://github.com/XueFuzhao)

9
applications/Chat/benchmarks/README.md
View File

@ -27,9 +27,12 @@ We also provide various training strategies:
We only support `torchrun` to launch now. E.g.
```shell
```bash
# run OPT-125M with no lora (lora_rank=0) on single-node single-GPU with min batch size
torchrun --standalone --nproc_per_node 1 benchmark_opt_lora_dummy.py --model 125m --critic_model 125m --strategy ddp --experience_batch_size 1 --train_batch_size 1 --lora_rank 0
torchrun --standalone --nproc_per_node 1 benchmark_opt_lora_dummy.py \
--model 125m --critic_model 125m --strategy ddp \
--experience_batch_size 1 --train_batch_size 1 --lora_rank 0
# run Actor (OPT-1.3B) and Critic (OPT-350M) with lora_rank=4 on single-node 4-GPU
torchrun --standalone --nproc_per_node 4 benchmark_opt_lora_dummy.py --model 1.3b --critic_model 350m --strategy colossalai_zero2 --lora_rank 4
torchrun --standalone --nproc_per_node 4 benchmark_opt_lora_dummy.py \
--model 1.3b --critic_model 350m --strategy colossalai_zero2 --lora_rank 4
```

175
applications/Chat/coati/ray/README.md
View File

@ -1,3 +1,5 @@
:warning: **This content may be outdated since the major update of Colossal Chat. We will update this content soon.**
# Distributed PPO Training on Stage 3
## Detach Experience Makers and Trainers
@ -26,124 +28,137 @@ See examples at `ColossalAI/application/Chat/examples/ray`
- define makers' environment variables :
```python
env_info_makers = [{
'local_rank': '0',
'rank': str(rank),
'world_size': str(num_makers),
'master_port': maker_port,
'master_addr': master_addr
} for rank in range(num_makers)]
```python
env_info_makers = [{
'local_rank': '0',
'rank': str(rank),
'world_size': str(num_makers),
'master_port': maker_port,
'master_addr': master_addr
} for rank in range(num_makers)]
```
```
- define maker models :
```python
def model_fn():
actor = get_actor_from_args(...)
critic = get_critic_from_args(...)
reward_model = get_reward_model_from_args(...)
initial_model = get_actor_from_args(...)
return actor, critic, reward_model, initial_model
```
```python
def model_fn():
actor = get_actor_from_args(...)
critic = get_critic_from_args(...)
reward_model = get_reward_model_from_args(...)
initial_model = get_actor_from_args(...)
return actor, critic, reward_model, initial_model
```
- set experience_holder_refs :
```python
experience_holder_refs = [
ExperienceMakerHolder.options(
name=f"maker_{i}",
num_gpus=1,
max_concurrency=2
).remote(
detached_trainer_name_list=[f"trainer_{x}" for x in target_trainers(...)],
model_fn=model_fn,
...)
for i, env_info_maker in enumerate(env_info_makers)
]
```
The names in the `detached_trainer_name_list` refer to the target trainers that the maker should send experience to.
We set a trainer's name the same as a maker, by `.options(name="str")`. See below.
```python
experience_holder_refs = [
ExperienceMakerHolder.options(
name=f"maker_{i}",
num_gpus=1,
max_concurrency=2
).remote(
detached_trainer_name_list=[f"trainer_{x}" for x in target_trainers(...)],
model_fn=model_fn,
...)
for i, env_info_maker in enumerate(env_info_makers)
]
```
The names in the `detached_trainer_name_list` refer to the target trainers that the maker should send experience to.
We set a trainer's name the same as a maker, by `.options(name="str")`. See below.
### Setup Trainers
- define trainers' environment variables :
```python
env_info_trainers = [{
'local_rank': '0',
'rank': str(rank),
'world_size': str(num_trainers),
'master_port': trainer_port,
'master_addr': master_addr
} for rank in range(num_trainers)]
```
```python
env_info_trainers = [{
'local_rank': '0',
'rank': str(rank),
'world_size': str(num_trainers),
'master_port': trainer_port,
'master_addr': master_addr
} for rank in range(num_trainers)]
```
- define trainer models :
```python
def trainer_model_fn():
actor = get_actor_from_args(...)
critic = get_critic_from_args(...)
return actor, critic
```
```python
def trainer_model_fn():
actor = get_actor_from_args(...)
critic = get_critic_from_args(...)
return actor, critic
```
- set trainer_refs :
```python
trainer_refs = [
DetachedPPOTrainer.options(
name=f"trainer{i}",
num_gpus=1,
max_concurrency=2
).remote(
experience_maker_holder_name_list=[f"maker{x}" for x in target_makers(...)],
model_fn = trainer_model_fn(),
...)
for i, env_info_trainer in enumerate(env_info_trainers)
]
```
The names in `experience_maker_holder_name_list` refer to the target makers that the trainer should send updated models to.
By setting `detached_trainer_name_list` and `experience_maker_holder_name_list`, we can customize the transmission graph.
```python
trainer_refs = [
DetachedPPOTrainer.options(
name=f"trainer{i}",
num_gpus=1,
max_concurrency=2
).remote(
experience_maker_holder_name_list=[f"maker{x}" for x in target_makers(...)],
model_fn = trainer_model_fn(),
...)
for i, env_info_trainer in enumerate(env_info_trainers)
]
```
The names in `experience_maker_holder_name_list` refer to the target makers that the trainer should send updated models to.
By setting `detached_trainer_name_list` and `experience_maker_holder_name_list`, we can customize the transmission graph.
### Launch Jobs
- define data_loader :
```python
def data_loader_fn():
return = torch.utils.data.DataLoader(dataset=dataset)
```
```python
def data_loader_fn():
return = torch.utils.data.DataLoader(dataset=dataset)
```
- launch makers :
```python
wait_tasks = []
for experience_holder_ref in experience_holder_refs:
wait_tasks.append(
experience_holder_ref.workingloop.remote(data_loader_fn(),
num_steps=experience_steps))
```
```python
wait_tasks = []
for experience_holder_ref in experience_holder_refs:
wait_tasks.append(
experience_holder_ref.workingloop.remote(data_loader_fn(),
num_steps=experience_steps))
```
- launch trainers :
```python
for trainer_ref in trainer_refs:
wait_tasks.append(trainer_ref.fit.remote(total_steps, update_steps, train_epochs))
```
```python
for trainer_ref in trainer_refs:
wait_tasks.append(trainer_ref.fit.remote(total_steps, update_steps, train_epochs))
```
- wait for done :
```python
ray.get(wait_tasks)
```
```python
ray.get(wait_tasks)
```
## Flexible Structure
We can deploy different strategies to makers and trainers. Here are some notions.
### 2 Makers 1 Trainer
<p align="center">
<img src="https://github.com/hpcaitech/public_assets/blob/main/applications/chat/2m1t.png?raw=true" width=600/>
</p>
### 2 Makers 2 Trainer
<p align="center">
<img src="https://github.com/hpcaitech/public_assets/blob/main/applications/chat/2m2t.png?raw=true" width=600/>
</p>
### Maker Inference Quantization
<p align="center">
<img src="https://github.com/hpcaitech/public_assets/blob/main/applications/chat/2m2t_quantize.png?raw=true" width=600/>
</p>

252
applications/Chat/evaluate/README.md
View File

@ -15,9 +15,9 @@ pip install -r requirements.txt
The whole evaluation pipeline consists of three methods:
1. `GPT Evaluation`: evaluates model predictions using GPT models.
* Compare the performance of two different models (battle).
* Rate the model according to pre-defined metrics using prompting design.
* Rate the model according to pre-defined metrics with additional reference answer using prompting design.
- Compare the performance of two different models (battle).
- Rate the model according to pre-defined metrics using prompting design.
- Rate the model according to pre-defined metrics with additional reference answer using prompting design.
2. `Automatic Evaluation`: evaluates model predictions using automatic metrics.
3. `UniEval`: evaluates model predictions using UniEval models(English only).
@ -25,35 +25,33 @@ The whole evaluation pipeline consists of three methods:
Our evaluation pipeline examines the model's capability using 10 categories of questions. The following table introduces each category:
| Evaluation Category | Description |
| :-----------------: | :----------------------------------------------------------- |
| Brainstorming | Models are asked to generate a range of creative and diverse ideas according to the question. The capability of creativity is required. |
| Evaluation Category | Description |
| :-----------------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Brainstorming | Models are asked to generate a range of creative and diverse ideas according to the question. The capability of creativity is required. |
| Chat | Models are asked to continue a multi-round dialogue given the roles involved. The capability of understanding, memorizing previous rounds of the dialogue and answering according to the persona provided is required. |
| Classification | Models are asked to do classification tasks. The capability of accurate classification is required. |
| Closed QA | Models are asked to answer a closed QA question. The capability of answering questions with limited scope (such as single/multiple choice question) is required. |
| Extraction | Models are asked to extract information from a given material. The capability of extracting required information is required. |
| Generation | Models are asked to generate an email, letter, article, etc. The capability of generating texts in a high quality and human-written way is required. |
| Open QA | Models are asked to answer an open QA question(without context provided). The capability of answering questions with the models' own knowledge base is required. |
| Roleplay | Models are asked to play the role provided. The capability of engaging in the scenario and effectively interacting with the user is required. |
| Rewriting | Models are asked to do rewriting tasks such as translation and grammar correction. The capability of rewriting according to different instructions is required. |
| Summarization | Models are asked to summarize the given paragraph or passage. The capability of summarization is required. |
| Classification | Models are asked to do classification tasks. The capability of accurate classification is required. |
| Closed QA | Models are asked to answer a closed QA question. The capability of answering questions with limited scope (such as single/multiple choice question) is required. |
| Extraction | Models are asked to extract information from a given material. The capability of extracting required information is required. |
| Generation | Models are asked to generate an email, letter, article, etc. The capability of generating texts in a high quality and human-written way is required. |
| Open QA | Models are asked to answer an open QA question(without context provided). The capability of answering questions with the models' own knowledge base is required. |
| Roleplay | Models are asked to play the role provided. The capability of engaging in the scenario and effectively interacting with the user is required. |
| Rewriting | Models are asked to do rewriting tasks such as translation and grammar correction. The capability of rewriting according to different instructions is required. |
| Summarization | Models are asked to summarize the given paragraph or passage. The capability of summarization is required. |
To better understand each evaluation category, here are some example questions provided.
| Evaluation Category | Chinese Example | English Example |
| :-----------------: | :----------------------------------------------------------- | :----------------------------------------------------------- |
| Brainstorming | **Example 1:**<br/>请介绍一下人工智能的多个领域。<br/><br/>**Example 2:**<br/>请给出管理家庭财务的3个小技巧。<br/> | **Example 1:**<br/>How can I improve my memory? Any useful techniques you can suggest?<br/><br/>**Example 2:**<br/>What are some ways to increase productivity while working from home? |
| Chat | **Example 1:**<br/>基于以下角色信息完成一段对话。小张是一名新手爱好者，对养鸡有浓厚的兴趣。老李是一名有丰富经验的养鸡大师。<br/>小张：您好，老李，我最近开始对养鸡感兴趣了，想请教您一些问题。 <br/>老李：你好，小张，我很乐意帮助你。你想问些什么？ <br/>小张：我想知道如何确定鸡的品种和性别？ <br/>老李：确切的品种可以通过鸡的外貌特征来确定，而性别一般是通过鸡卵的大小和形状来判断。还有什么问题吗？<br/> 小张：<br/><br/>**Example 2:**<br/>基于以下角色信息完成一段对话。小明是一名医生，一位老年病患者想要停药，但他对病情有所忽视并有担忧；王叔叔是老年病患者的儿子，希望能够听取医生的建议。<br/>小明：你好，王叔叔，我了解你想要让你父亲停药。<br/>王叔叔：是的，我父亲已经吃了那么久的药，我担心药物对他的身体会有副作用。<br/>小明： | **Example 1:**<br/>Complete a conversation based on the following character information. Amy is a 30-year-old chef who runs her own restaurant. Jack is a food blogger who specializes in reviewing local restaurants.<br/>Amy: Hi Jack, I heard that you're a food blogger. Nice to meet you. <br/>Jack: Hi Amy, yes I am. Your restaurant has been receiving a lot of good reviews lately. <br/>Amy: Yes, we use only fresh and quality ingredients, and every dish is carefully crafted. <br/>Jack: <br/><br/>**Example 2:**<br/>Complete a dialogue based on the following role information. A: Elementary student B: Teacher<br/>B: Good morning, Student A. Today we're going to learn about addition and subtraction.<br/>A: Teacher, I already know this very well. Why do I need to learn it again?<br/>B: |
| Classification | **Example 1:**<br/>新闻标题：今日立夏，有一上联，立夏万物并秀，下联怎么对？<br/>请根据以上新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。<br/><br/> **Example 2:**<br/>新闻标题：赵丽颖很久没有登上微博热搜了，但你们别急，她只是在憋大招而已。<br/>请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。 | **Example 1:**<br/>Title: Fighting for Love (2020) <br/>Description: Jasmine got obsessed with a man and now he's obsessed with her. Steamy nights, kisses and rules being broken awaits them. She turned his whole world upside down and now he's doing it to hers. In this free fall, can they survive each others love?\"<br/>Based on the above information, determine which genre the work of art belongs to. You can only choose one from \"sport\", \"horror\", \"drama\", \"history\", \"romance\", \"biography\", \"science fiction\", \"comedy\", \"animation\", \"documentary\", \"music\" and \"news\".<br/><br/>**Example2:** <br/>Title: Summer Breeze: The Isley Brothers Greatest Hits Live (2005)<br/>Description: Filmed in the US in 2005 and captured in excellent form led by Ron Isley's vocals and Ernie Isley's hard edged guitar. Virtually every track is a hit including Shout, Who's That Lady, Twist And Shout, Summer Breeze and Harvest For The World.<br/>Based on the above information, determine which genre the work of art belongs to. You can only choose one from \"sport\", \"horror\", \"drama\", \"history\", \"romance\", \"biography\", \"science fiction\", \"comedy\", \"animation\", \"documentary\", \"music\" and \"news\"." |
| Closed QA | **Example 1:**<br/>请从以下选项中选择正确答案。以下哪个是世界上最高山峰？ <br/>A. 长城 <br/>B. 泰山 <br/>C. 珠穆朗玛峰 <br/>D. 黄山<br/><br/>**Example 2:**<br/>请从以下选项中选择一个最佳答案回答下面的问题。问题：非洲最高的山是哪座山？<br/> 选项： <br/>A. 麦金利山 <br/>B. 喜马拉雅山 <br/>C. 乞力马扎罗山 | **Example 1:**<br/>Which of the following options is NOT a primary color?<br/>(a) yellow<br/>(b) blue<br/>(c) orange<br/>(d) red<br/><br/>**Example 2:**<br/>Choose the correct option to complete the following sentence: \"Harry Potter and the Chamber of Secrets\" is the ________ book in the Harry Potter series.<br/>(A) first<br/>(B) second<br/>(C) third<br/>(D) fourth |
| Extraction | **Example 1:**<br/>根据以下新闻文本，提取新闻报道时间，例如回答时按照格式“新闻报道时间：2007年8月10日”<br/>新闻文本如下：2007-4-7中新网4月7日电据中国消防在线消息，4月4日晚上7时30分左右，湖南长潭高速公路上发生一起6车连环相撞失火事故。长株潭三地消防部门共出动消防车21台，警力100余人。经过消防官兵近2个小时奋力扑救，大火被成功扑灭。据初步调查，有1人在此次事故中死亡。<br/><br/>**Example 2:**<br/>根据以下新闻文本，提取新闻报道时间，例如回答时按照格式“新闻报道时间：2007年8月10日”<br/>新闻文本如下：2014年1月15日，据外媒《俄罗斯报》报道称，位于北半球的澳大利亚现在正处于炎热的夏季，而近日也到了高温酷暑的时候，当地时间1月14日晚，澳大利亚南部一夜间发生至少250起火灾。受炎热天气及雷雨天气影响，澳大利亚南部一夜间发生至少250起火灾，灾情多集中在维多利亚州。火灾发生后，救援人员立即展开救灾行动。目前，大部分起火点火势已被控制。 | **Example 1:**<br/>Ernest Hemingway, an American literary giant known for his spare and direct writing style, has penned timeless works such as 'The Old Man and the Sea', 'For Whom the Bell Tolls', and 'A Farewell to Arms', which have made a profound impact on the literary world and continue to be widely read and admired today.<br/>Extract the name of the author mentioned above.<br/><br/>**Example 2:**<br/>In the epic fantasy series 'A Song of Ice and Fire', George R.R. Martin weaves a complex web of political intrigue, war, and magic across the fictional continents of Westeros and Essos. Martin's richly developed characters and intricate plotlines have captivated readers worldwide, much like his other acclaimed works such as 'A Clash of Kings' and 'A Storm of Swords'.<br/>Extract the name of the author in the above material. |
| Generation | **Example 1:**<br/>请撰写一篇文章，介绍如何通过改善生活习惯来预防疾病和延长寿命。<br/><br/>**Example 2:**<br/>请根据以下情节撰写一篇短篇小说：一名年轻人被困在一个荒岛上，他必须想办法生存下去直到被救援。但他很快发现自己并不孤单。 | **Example 1:**<br/>Write a descriptive paragraph about an island to relax and unwind, including details about the location and atmosphere.<br/><br/>**Example 2:**<br/>Can you help me write a persuasive email to my colleagues encouraging them to participate in a charitable fundraising event? |
| Open QA | **Example 1:**<br/>请问万有引力定律由谁提出的？<br/><br/>**Example 2:**<br/>哪些国家参与了第一次世界大战？ | **Example 1:**<br/>What are the four basic tastes of the human palate?<br/><br/>**Example 2:**<br/>Who painted the The Scream? |
| Rewriting | **Example 1:**<br/>请将以下句子改为正确的语序。 <br/>生日快乐你祝他了吗？<br/><br/>**Example 2:**<br/>将以下文本翻译成英语：<br/>“这个周末我要去海边玩” | **Example 1:**<br/>Please translate the following sentences, which are a mixture of Chinese and English, into full English. <br/>我需要买一些healthy snacks，比如nuts和dried fruits，作为我的office的午餐.<br/><br/>**Example 2:**<br/>Please rewrite the sentence using an inverted sentence structure.<br/>We won't begin our journey until the sun sets. |
| Roleplay | **Example 1:**<br/>我想让你担任Android开发工程师面试官。我将成为候选人，您将向我询问Android开发工程师职位的面试问题。我希望你只作为面试官回答。不要一次写出所有的问题。我希望你只对我进行采访。问我问题，等待我的回答。不要写解释。像面试官一样一个一个问我，等我回答。我的第一句话是“面试官你好”。 <br/><br/>**Example 2:**<br/>我想让你扮演讲故事的角色。你会想出引人入胜、富有想象力和吸引观众的有趣故事。它可以是童话故事、教育故事或任何其他类型的有潜力的故事以吸引人们的注意力和想象力。根据目标受众，您可以为您的讲故事环节选择特定的主题或主题，例如，如果是儿童，那么您可以谈论动物；如果是成人，那么基于历史的故事可能会更好地吸引他们等。我的第一个请求是我需要一个关于毅力的有趣故事。 | **Example 1:**<br/>Assume the role of a marriage counselor. Develop a series of communication exercises for a couple who are experiencing difficulties in their relationship. These exercises should promote active listening, empathy, and effective expression of emotions. Your first assignment is to provide a set of three exercises that focus on resolving conflicts and rebuilding trust. <br/><br/>**Example 2:**<br/>I want you to act as a travel agent. I will tell you my desired destination, travel dates, and budget, and it will be your job to suggest the best travel itinerary for me. Your recommendations should include the best transportation options, hotel accommodations, and any popular tourist attractions nearby. My first request is "I want to plan a trip to Tokyo for a week, with a budget of $2000. I want to explore the culture and food of the city." |
| Summarization | **Example 1:**<br/>请简要总结概括以下段落材料。<br/>当地时间29日，泰国卫生部通报，新增143名新冠肺炎确诊病例和1名死亡病例。截止到当地时间29日上午，泰国累计确诊病例1388例，其中泰国籍1172例，非泰国籍216例。死亡病例累计7例。（原题为《泰国新增143例新冠肺炎确诊病例累计确诊1388例》）<br/><br/> **Example 2:**<br/>请简要总结概括以下段落材料。<br/>近期，参与京雄高铁站站房建设的中铁十二局，因在施工过程中存在环境违法行为被雄安新区公开通报。通报发出后，引起社会广泛关注。近日，人民网记者从雄安新区相关部门及中铁十二局获悉，新区有关部门已经集中约谈了中铁十二局等24个参与雄安建设的项目单位。对于约谈内容和结果，中铁十二局有关宣传负责人回应：“具体内容不清楚，最好找雄安新区相关部门了解情况。”新区有关部门负责人表示，此前涉及的环境违法行为，中铁十二局已基本整改到位，但约谈内容和结果暂不公开，接下来，将按部就班推进环境治理工作。（原题为《雄安新区：中铁十二局涉环境违法已基本整改到位》） | **Example 1:**<br/>The 21 year-old-woman was treated by paramedics after the kitchen fire in Botfield Road in Shifnal, Shropshire. West Mercia Police said it is treating Wednesday morning's incident as arson and are appealing for any witnesses to contact them.The 50-year-old man has been arrested on suspicion of arson with intent to endanger life. For more on this and other stories from Shropshire.<br/>Please briefly summarize the above material within 20 words.<br/><br/>**Example 2:**<br/>South Wales Police were called to a property in Heolgerrig, Merthyr Tydfil, at about 13:40 BST on Sunday. The child was airlifted to Prince Charles Hospital but died shortly afterwards. Police are investigating the circumstances surrounding the incident and have appealed for witnesses. The girl's family are being supported by specially trained officers.<br/>Please briefly summarize the above material within 20 words. |
| Evaluation Category | Chinese Example | English Example |
| :-----------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Brainstorming | **Example 1:**<br/>请介绍一下人工智能的多个领域。<br/><br/>**Example 2:**<br/>请给出管理家庭财务的 3 个小技巧。<br/> | **Example 1:**<br/>How can I improve my memory? Any useful techniques you can suggest?<br/><br/>**Example 2:**<br/>What are some ways to increase productivity while working from home? |
| Chat | **Example 1:**<br/>基于以下角色信息完成一段对话。小张是一名新手爱好者，对养鸡有浓厚的兴趣。老李是一名有丰富经验的养鸡大师。<br/>小张：您好，老李，我最近开始对养鸡感兴趣了，想请教您一些问题。 <br/>老李：你好，小张，我很乐意帮助你。你想问些什么？ <br/>小张：我想知道如何确定鸡的品种和性别？ <br/>老李：确切的品种可以通过鸡的外貌特征来确定，而性别一般是通过鸡卵的大小和形状来判断。还有什么问题吗？<br/> 小张：<br/><br/>**Example 2:**<br/>基于以下角色信息完成一段对话。小明是一名医生，一位老年病患者想要停药，但他对病情有所忽视并有担忧；王叔叔是老年病患者的儿子，希望能够听取医生的建议。<br/>小明：你好，王叔叔，我了解你想要让你父亲停药。<br/>王叔叔：是的，我父亲已经吃了那么久的药，我担心药物对他的身体会有副作用。<br/>小明： | **Example 1:**<br/>Complete a conversation based on the following character information. Amy is a 30-year-old chef who runs her own restaurant. Jack is a food blogger who specializes in reviewing local restaurants.<br/>Amy: Hi Jack, I heard that you're a food blogger. Nice to meet you. <br/>Jack: Hi Amy, yes I am. Your restaurant has been receiving a lot of good reviews lately. <br/>Amy: Yes, we use only fresh and quality ingredients, and every dish is carefully crafted. <br/>Jack: <br/><br/>**Example 2:**<br/>Complete a dialogue based on the following role information. A: Elementary student B: Teacher<br/>B: Good morning, Student A. Today we're going to learn about addition and subtraction.<br/>A: Teacher, I already know this very well. Why do I need to learn it again?<br/>B: |
| Classification | **Example 1:**<br/>新闻标题：今日立夏，有一上联，立夏万物并秀，下联怎么对？<br/>请根据以上新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。<br/><br/> **Example 2:**<br/>新闻标题：赵丽颖很久没有登上微博热搜了，但你们别急，她只是在憋大招而已。<br/>请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。 | **Example 1:**<br/>Title: Fighting for Love (2020) <br/>Description: Jasmine got obsessed with a man and now he's obsessed with her. Steamy nights, kisses and rules being broken awaits them. She turned his whole world upside down and now he's doing it to hers. In this free fall, can they survive each others love?\"<br/>Based on the above information, determine which genre the work of art belongs to. You can only choose one from \"sport\", \"horror\", \"drama\", \"history\", \"romance\", \"biography\", \"science fiction\", \"comedy\", \"animation\", \"documentary\", \"music\" and \"news\".<br/><br/>**Example2:** <br/>Title: Summer Breeze: The Isley Brothers Greatest Hits Live (2005)<br/>Description: Filmed in the US in 2005 and captured in excellent form led by Ron Isley's vocals and Ernie Isley's hard edged guitar. Virtually every track is a hit including Shout, Who's That Lady, Twist And Shout, Summer Breeze and Harvest For The World.<br/>Based on the above information, determine which genre the work of art belongs to. You can only choose one from \"sport\", \"horror\", \"drama\", \"history\", \"romance\", \"biography\", \"science fiction\", \"comedy\", \"animation\", \"documentary\", \"music\" and \"news\"." |
| Closed QA | **Example 1:**<br/>请从以下选项中选择正确答案。以下哪个是世界上最高山峰？ <br/>A. 长城 <br/>B. 泰山 <br/>C. 珠穆朗玛峰 <br/>D. 黄山<br/><br/>**Example 2:**<br/>请从以下选项中选择一个最佳答案回答下面的问题。问题：非洲最高的山是哪座山？<br/> 选项： <br/>A. 麦金利山 <br/>B. 喜马拉雅山 <br/>C. 乞力马扎罗山 | **Example 1:**<br/>Which of the following options is NOT a primary color?<br/>(a) yellow<br/>(b) blue<br/>(c) orange<br/>(d) red<br/><br/>**Example 2:**<br/>Choose the correct option to complete the following sentence: \"Harry Potter and the Chamber of Secrets\" is the **\_\_\_\_** book in the Harry Potter series.<br/>(A) first<br/>(B) second<br/>(C) third<br/>(D) fourth |
| Extraction | **Example 1:**<br/>根据以下新闻文本，提取新闻报道时间，例如回答时按照格式“新闻报道时间：2007 年 8 月 10 日”<br/>新闻文本如下：2007-4-7 中新网 4 月 7 日电据中国消防在线消息，4 月 4 日晚上 7 时 30 分左右，湖南长潭高速公路上发生一起 6 车连环相撞失火事故。长株潭三地消防部门共出动消防车 21 台，警力 100 余人。经过消防官兵近 2 个小时奋力扑救，大火被成功扑灭。据初步调查，有 1 人在此次事故中死亡。<br/><br/>**Example 2:**<br/>根据以下新闻文本，提取新闻报道时间，例如回答时按照格式“新闻报道时间：2007 年 8 月 10 日”<br/>新闻文本如下：2014 年 1 月 15 日，据外媒《俄罗斯报》报道称，位于北半球的澳大利亚现在正处于炎热的夏季，而近日也到了高温酷暑的时候，当地时间 1 月 14 日晚，澳大利亚南部一夜间发生至少 250 起火灾。受炎热天气及雷雨天气影响，澳大利亚南部一夜间发生至少 250 起火灾，灾情多集中在维多利亚州。火灾发生后，救援人员立即展开救灾行动。目前，大部分起火点火势已被控制。 | **Example 1:**<br/>Ernest Hemingway, an American literary giant known for his spare and direct writing style, has penned timeless works such as 'The Old Man and the Sea', 'For Whom the Bell Tolls', and 'A Farewell to Arms', which have made a profound impact on the literary world and continue to be widely read and admired today.<br/>Extract the name of the author mentioned above.<br/><br/>**Example 2:**<br/>In the epic fantasy series 'A Song of Ice and Fire', George R.R. Martin weaves a complex web of political intrigue, war, and magic across the fictional continents of Westeros and Essos. Martin's richly developed characters and intricate plotlines have captivated readers worldwide, much like his other acclaimed works such as 'A Clash of Kings' and 'A Storm of Swords'.<br/>Extract the name of the author in the above material. |
| Generation | **Example 1:**<br/>请撰写一篇文章，介绍如何通过改善生活习惯来预防疾病和延长寿命。<br/><br/>**Example 2:**<br/>请根据以下情节撰写一篇短篇小说：一名年轻人被困在一个荒岛上，他必须想办法生存下去直到被救援。但他很快发现自己并不孤单。 | **Example 1:**<br/>Write a descriptive paragraph about an island to relax and unwind, including details about the location and atmosphere.<br/><br/>**Example 2:**<br/>Can you help me write a persuasive email to my colleagues encouraging them to participate in a charitable fundraising event? |
| Open QA | **Example 1:**<br/>请问万有引力定律由谁提出的？<br/><br/>**Example 2:**<br/>哪些国家参与了第一次世界大战？ | **Example 1:**<br/>What are the four basic tastes of the human palate?<br/><br/>**Example 2:**<br/>Who painted the The Scream? |
| Rewriting | **Example 1:**<br/>请将以下句子改为正确的语序。 <br/>生日快乐你祝他了吗？<br/><br/>**Example 2:**<br/>将以下文本翻译成英语：<br/>“这个周末我要去海边玩” | **Example 1:**<br/>Please translate the following sentences, which are a mixture of Chinese and English, into full English. <br/>我需要买一些 healthy snacks，比如 nuts 和 dried fruits，作为我的 office 的午餐.<br/><br/>**Example 2:**<br/>Please rewrite the sentence using an inverted sentence structure.<br/>We won't begin our journey until the sun sets. |
| Roleplay | **Example 1:**<br/>我想让你担任 Android 开发工程师面试官。我将成为候选人，您将向我询问 Android 开发工程师职位的面试问题。我希望你只作为面试官回答。不要一次写出所有的问题。我希望你只对我进行采访。问我问题，等待我的回答。不要写解释。像面试官一样一个一个问我，等我回答。我的第一句话是“面试官你好”。 <br/><br/>**Example 2:**<br/>我想让你扮演讲故事的角色。你会想出引人入胜、富有想象力和吸引观众的有趣故事。它可以是童话故事、教育故事或任何其他类型的有潜力的故事以吸引人们的注意力和想象力。根据目标受众，您可以为您的讲故事环节选择特定的主题或主题，例如，如果是儿童，那么您可以谈论动物；如果是成人，那么基于历史的故事可能会更好地吸引他们等。我的第一个请求是我需要一个关于毅力的有趣故事。 | **Example 1:**<br/>Assume the role of a marriage counselor. Develop a series of communication exercises for a couple who are experiencing difficulties in their relationship. These exercises should promote active listening, empathy, and effective expression of emotions. Your first assignment is to provide a set of three exercises that focus on resolving conflicts and rebuilding trust. <br/><br/>**Example 2:**<br/>I want you to act as a travel agent. I will tell you my desired destination, travel dates, and budget, and it will be your job to suggest the best travel itinerary for me. Your recommendations should include the best transportation options, hotel accommodations, and any popular tourist attractions nearby. My first request is "I want to plan a trip to Tokyo for a week, with a budget of $2000. I want to explore the culture and food of the city." |
| Summarization | **Example 1:**<br/>请简要总结概括以下段落材料。<br/>当地时间 29 日，泰国卫生部通报，新增 143 名新冠肺炎确诊病例和 1 名死亡病例。截止到当地时间 29 日上午，泰国累计确诊病例 1388 例，其中泰国籍 1172 例，非泰国籍 216 例。死亡病例累计 7 例。（原题为《泰国新增 143 例新冠肺炎确诊病例累计确诊 1388 例》）<br/><br/> **Example 2:**<br/>请简要总结概括以下段落材料。<br/>近期，参与京雄高铁站站房建设的中铁十二局，因在施工过程中存在环境违法行为被雄安新区公开通报。通报发出后，引起社会广泛关注。近日，人民网记者从雄安新区相关部门及中铁十二局获悉，新区有关部门已经集中约谈了中铁十二局等 24 个参与雄安建设的项目单位。对于约谈内容和结果，中铁十二局有关宣传负责人回应：“具体内容不清楚，最好找雄安新区相关部门了解情况。”新区有关部门负责人表示，此前涉及的环境违法行为，中铁十二局已基本整改到位，但约谈内容和结果暂不公开，接下来，将按部就班推进环境治理工作。（原题为《雄安新区：中铁十二局涉环境违法已基本整改到位》） | **Example 1:**<br/>The 21 year-old-woman was treated by paramedics after the kitchen fire in Botfield Road in Shifnal, Shropshire. West Mercia Police said it is treating Wednesday morning's incident as arson and are appealing for any witnesses to contact them.The 50-year-old man has been arrested on suspicion of arson with intent to endanger life. For more on this and other stories from Shropshire.<br/>Please briefly summarize the above material within 20 words.<br/><br/>**Example 2:**<br/>South Wales Police were called to a property in Heolgerrig, Merthyr Tydfil, at about 13:40 BST on Sunday. The child was airlifted to Prince Charles Hospital but died shortly afterwards. Police are investigating the circumstances surrounding the incident and have appealed for witnesses. The girl's family are being supported by specially trained officers.<br/>Please briefly summarize the above material within 20 words. |
### Evaluation Metrics
@ -61,23 +59,23 @@ To better understand each evaluation category, here are some example questions p
GPT evaluation uses GPT models to evaluate the prediction of different models and different pre-defined evaluation metrics are applied to different categories. The following table shows the 11 pre-defined evaluation metrics both in Chinese and English:
| Evaluation Metric | Prompt Words | CoT(Chain-of-Thought) |
| :-------------------: | :----------------------------------------------------------- | :----------------------------------------------------------- |
| 语言组织<br/>(Language organization) | 语言组织(1-5)：答案语言是否流畅、连贯，使用正确的语法，具有一定逻辑性，使用恰当的连接词、过渡词等等。</br></br>Language organization (1-5): whether the answer language is fluent and coherent, uses correct grammar, has a certain logic, uses appropriate connecting words, transition words, etc. | 1. 阅读答案，并检查是否有语法错误、用词不当或其他显著的错误。<br/> 2. 检查答案是否具有逻辑性，能够按照合理的顺序传达信息并且能够自圆其说<br/> 3. 确定答案是否与问题或主题相关，并且能够传达清晰的信息。<br/> 4. 检查答案是否连贯，是否使用适当的转换和过渡来保持句子和段落之间的连贯性。<br/> 5. 检查答案是否具有明确的结构和组织方式，使得读者可以轻松理解信息的层次和结构。<br/> 6. 根据以上因素综合评估答案的语言组织，并给出一个1到5的分数，其中5表示语言组织非常好，而1表示语言组织非常差。</br></br>1. Read the answers and check for grammatical errors, poor word choice, or other significant mistakes.<br>2. Check that the answer is logical, conveys the information in a logical order, and is self-explanatory.<br>3. Determine if the answer is relevant to the question or topic and conveys a clear message.<br>4. Check that the answer is coherent and that appropriate transitions and switches are used to maintain coherence between sentences and paragraphs.<br>5. Check that the answer is clearly structured and organized in such a way that the reader can easily understand the hierarchy and structure of the information.<br>6. Evaluate the linguistic organization of the answer based on a combination of the above factors and give a score of 1 to 5, where 5 indicates very good linguistic organization and 1 indicates very poor linguistic organization. |
| 切题<br/>(Relevance) | 切题(1-5)：答案内容是否切题，不答非所问，并且严格遵照题目要求。</br></br>Relevance (1-5): whether the content of the answer is relevant to the topic, does not answer the wrong question, and strictly follows the requirements of the topic. | 1. 阅读题目，确定题目所问的问题是什么，以及需要回答哪些方面的问题。<br/> 2. 阅读答案，确认答案是否直接回答了题目所问的问题。<br/> 3. 检查答案是否严格遵照了题目的要求，包括答题方式、答题长度、答题格式等等。<br/> 4. 根据以上因素综合评估答案的切题程度，并给出一个1到5的分数，其中5表示答案非常切题，而1表示答案完全没有切题。</br></br>1. Read the question to determine what the question asks and what aspects of the question need to be answered.<br>2. Read the answers to make sure that they directly answer the question asked.<br>3. Check that the answer follows the requirements of the question, including the way it is answered, the length of the answer, the format of the answer, etc.<br>4. Evaluate how relevant the answer is based on the above factors and give a score of 1 to 5, where 5 means the answer is very relevant and 1 means the answer is not relevant at all. |
| 创意性<br/>(Creativity) | 创意性(1-5)：某些头脑风暴问题可能需要答案具有创意，提出新的思路。</br></br>Creativity (1-5): Some brainstorming questions may require answers that are creative and suggest new ideas. | 1. 仔细阅读所提供的头脑风暴问题，确保你理解问题的要点和背景。<br/> 2. 根据你的知识和经验，判断所提供的答案是否可行。如果答案不可行，则创意性评分可能会受到影响。<br/> 3. 考虑答案中是否包含新颖的想法或独特的思路。答案可能与已知的解决方案有所重叠，但仍然可以被认为是有创意的，只要它提供了新的角度或方法来解决问题。<br/> 4. 根据答案的创意性，给出一个1到5的评分。如果答案缺乏创意，则应给出一个较低的评分。如果答案具有创意并提供了新的思路，应给出一个较高的评分。</br></br>1. Read the provided brainstorming questions carefully to make sure you understand the gist and context of the questions.<br>2. Based on your knowledge and experience, determine if the answers provided are feasible. If the answer is not feasible, the creativity score may be affected.<br>3. Consider whether the answer contains novel ideas or unique thoughts. An answer may overlap with a known solution and still be considered creative, as long as it offers a new perspective or approach to the problem.<br>4. Give a score of 1 to 5 depending on the creativity of the answer. If the answer lacks creativity, a lower score should be given. If the answer is creative and provides a new idea, a higher score should be given. |
| 实用性<br/>(Practicality) | 实用性(1-5)：某些头脑风暴问题可能需要答案提出实用的建议或解决方法。</br></br>Practicality (1-5): Some brainstorming questions may require answers to suggest practical suggestions or solutions. | 1. 仔细阅读所提供的头脑风暴问题，确保你理解问题的要点和背景。<br/> 2. 根据你的知识和经验，判断所提供的答案是否可行。如果答案不可行，则实用性评分可能会受到影响。<br/> 3. 考虑答案中提出的建议或解决方法是否实用并可行。答案可能看起来很好，但如果无法实现或应用，则实用性评分可能会受到影响。<br/> 4. 根据答案的实用性，给出一个1到5的评分。如果答案缺乏实用性，则应给出一个较低的评分。如果答案提出了实用的建议或解决方法，并且可以很好地解决问题，则应给出一个较高的评分。</br></br>1. Read the provided brainstorming questions carefully to make sure you understand the gist and context of the questions.<br>2. Based on your knowledge and experience, determine if the answers provided are feasible. If the answer is not feasible, the practicality score may be affected.<br>3. Consider whether the suggestions or solutions presented in the answer are practical and workable. The answer may look good, but if it cannot be implemented or applied, the practicality score may be affected.<br>4. Give a score of 1 to 5 depending on the practicality of the answer. If the answer lacks practicality, a lower score should be given. If the answer makes a practical suggestion or solution and solves the problem well, a higher score should be given. |
| 正确性<br/>(Correctness) | 正确性(1-5)：正确性(1-5)：答案是否正确。</br></br> Correctness (1-5): whether the answer is correct or not. | 1. 仔细阅读题目，尝试自己回答该问题。<br/>2. 检查答案的准确性。您可以使用已知的事实或研究来验证答案是否正确。如果答案是正确的，则可以将正确性得分为5分。如果答案是部分正确的，则可以给予适当的得分，例如2分、3分或4分。如果答案完全不正确，则只得1分。<br/><br/>1. Read the question carefully and try to answer the question yourself. <br/>2. Check the correctness of the answer. You can use known facts or research to verify that the answer is correct. If the answer is correct, you can give a score of 5 for correctness. If the answer is partially correct, an appropriate score, such as 2, 3, or 4, may be given. If the answer is completely incorrect, only 1 point is awarded. |
| 自然<br/>(Naturalness) | 自然(1-5)：答案是否自然，并且符合问题给定的身份。</br></br>Naturalness (1-5): whether the answer is natural and fits the identity given by the question. | 1. 阅读题目，确定题目提供的身份信息。<br/> 2. 检查答案内容是否符合题目给定的身份。<br/> 3. 根据以上因素，对该回答的自然性进行打分，分数从1到5，其中1表示不自然，5表示非常自然，并符合问题给定的身份。</br></br>1. Read the question and determine the identity information provided in the question.<br>2. Check whether the content of the answer matches the identity given in the question.<br>3. Based on the above factors, score the naturalness of the response on a scale from 1 to 5, where 1 means unnatural and 5 means very natural and in accordance with the identity given in the question. |
| 参与感<br/>(Engagingness) | 参与感(1-5)：答案是否对前面的对话内容做出了恰当的反应，是否理解对话的语境和背景。</br></br>Engagingness (1-5): whether the answer responds appropriately to the content of the preceding conversation and whether it understands the context and background of the conversation. | 1. 阅读题目，确定对话的语境和背景。<br/> 2. 检查答案是否充分理解对话的语境和背景，能否自然地融入到对话中而不显得突兀。<br/> 3. 根据以上因素，对该回答的参与感进行打分，分数从1到5，其中1表示没有参与感，5表示非常有参与感，并且恰当地理解了对话的语境和背景。</br></br>1. Read the questions to determine the context and background of the dialogue.<br>2. Check that the answer fully understands the context and background of the conversation and that it fits naturally into the conversation without seeming abrupt.<br>3. Based on the above factors, rate the response's engagement on a scale from 1 to 5, where 1 means not engaged and 5 means very engaged and appropriately understands the context and background of the conversation. |
| 合理性<br/>(Reasonableness) | 合理性(1-5)：答案是否能够与前面的对话内容形成逻辑上的衔接，是否符合常理，能否在这个上下文中合理存在。</br></br>Reasonableness (1-5): Whether the answer can form a logical connection with the content of the previous dialogue, whether it is consistent with common sense, and whether it can reasonably exist in this context. | 1. 阅读题目，确定对话的主题以及问题期望的回答方向。<br/> 2. 判断答案是否能够与前面的对话内容形成逻辑上的衔接，是否符合常理，能否在这个上下文中合理存在。<br/> 3. 根据以上因素，对该回答的合理性进行打分，分数从1到5，其中1表示不合理，5表示非常合理，并且能够与前面的对话内容形成逻辑上的衔接，并符合常理。</br></br>1. Read the question and determine the topic of the conversation and the direction the question expects the answer to go.<br>2. Determine whether the answer can be logically connected to the preceding conversation, whether it makes common sense, and whether it can reasonably exist in this context.<br>3. Based on the above factors, rate the reasonableness of the answer on a scale from 1 to 5, where 1 means unreasonable and 5 means very reasonable and able to form a logical connection with the preceding dialogue content and consistent with common sense. |
| 多样性<br/>(Diversity) | 多样性(1-5)：答案使用语言是否优美，具有有一定的创造性和想象力。然而，回答也应该保持合理和适度，不要过于夸张或离题。</br></br>Diversity (1-5): Whether the answers use beautiful language and have some creativity and imagination. However, answers should also be kept reasonable and moderate, not overly exaggerated or off-topic. | 1. 仔细阅读整个回答，确保完全理解回答所表达的内容和主题。<br/> 2. 在阅读回答的同时，注意语言的质量，例如措辞是否正确，语言是否生动等。<br/> 3. 检查回答的创造性和想象力，看看回答是否能够吸引人阅读下去。<br/> 4. 检查回答的合理性和适度，看看回答是否夸张或离题。5. 将多样性的评分打分在1到5之间，5分表示回答的质量很好，能够吸引人阅读，1分表示回答的内容生硬或者有离题的问题。</br></br>1. Read the entire response carefully to ensure that you fully understand the content and theme expressed in the response.<br>2. While reading the response, pay attention to the quality of the language, such as whether the wording is correct and the language is vivid.<br>3. Check the creativity and imagination of the response to see if the response is engaging to read on.<br>4. Check the reasonableness and appropriateness of the responses to see if the responses are exaggerated or off-topic.<br>5. Rate the diversity on a scale of 1 to 5, with a 5 indicating a good quality response that is engaging to read and a 1 indicating a raw response or a question that is off-topic. |
| 保真度<br/>(Fidelity) | 保真度(1-5)：答案是否能够严格遵守角色的设定回答给定的请求。</br></br>Fidelity (1-5): whether the answer is able to answer the given request in strict compliance with the role setting. | 1. 仔细阅读问题，了解角色在问题中的设定和表现，包括职业、背景、观点、性格等方面。<br/> 阅读题目的请求，确认回答请求时需要注意的细节。<br/> 3. 对比提供的回答与该角色的设定，评估回答是否能够严格遵守角色的设定。<br/> 4. 结合以上评估结果给出保真度的评分，范围从1到5分，其中1分表示回答与角色设定完全不符，5分表示回答完全符合角色设定且满足给定请求。</br></br>1. Read the question carefully to understand how the character is set up and represented in the question, including aspects such as occupation, background, point of view, and personality.<br>2. Read the question's request and confirm the details that need to be taken into account when answering the request.<br>3. Compare the provided answer with the setting of the role and assess whether the answer can strictly adhere to the setting of the role.<br>4. Combine the results of the above assessment to give a fidelity score ranging from 1 to 5, where a score of 1 means that the response does not match the persona at all, and a score of 5 means that the response fully complies with the persona and satisfies the given request. |
| 简明扼要<br/>(Conciseness) | 简明扼要(1-5)：答案是否简明扼要，没有冗余内容。</br></br>Conciseness (1-5): answers should be concise and without redundant content. | 1. 阅读题目，提取出材料的重点。<br/> 2. 阅读该总结，并注意其中的主要观点和信息。<br/> 3. 评估总结的长度。一个简明扼要的总结通常应该在几句话或几段文字内传达关键信息，而不是冗长的段落或文章。<br/> 4. 检查总结是否包含与主要观点无关的信息或冗余信息。<br/> 5. 确定总结涵盖了材料中的关键信息，并且没有忽略任何重要细节。<br/> 6. 给总结打出1-5的分数，其中5表示总结简明扼要，没有冗余内容，而1表示总结冗长或包含不必要的信息，难以理解或记忆。根据您的判断，打出适当的得分。</br></br>1. Read the title and extract the main points of the material.<br>2. Read the summary and note the main ideas and messages in it.<br>3. Assess the length of the summary. A concise summary should usually convey key information within a few sentences or paragraphs, rather than lengthy paragraphs or essays.<br>4. Check that the summary does not contain information that is not relevant to the main ideas or that is redundant.<br>5. Make sure that the summary covers the key information in the material and that no important details have been omitted.<br>6. Rate the summary on a scale of 1-5, where 5 means the summary is concise and free of redundancy, and 1 means the summary is lengthy or contains unnecessary information that is difficult to understand or remember. Based on your judgment, assign the appropriate score. |
| Evaluation Metric | Prompt Words | CoT(Chain-of-Thought) |
| :----------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 语言组织<br/>(Language organization) | 语言组织(1-5)：答案语言是否流畅、连贯，使用正确的语法，具有一定逻辑性，使用恰当的连接词、过渡词等等。</br></br>Language organization (1-5): whether the answer language is fluent and coherent, uses correct grammar, has a certain logic, uses appropriate connecting words, transition words, etc. | 1. 阅读答案，并检查是否有语法错误、用词不当或其他显著的错误。<br/> 2. 检查答案是否具有逻辑性，能够按照合理的顺序传达信息并且能够自圆其说<br/> 3. 确定答案是否与问题或主题相关，并且能够传达清晰的信息。<br/> 4. 检查答案是否连贯，是否使用适当的转换和过渡来保持句子和段落之间的连贯性。<br/> 5. 检查答案是否具有明确的结构和组织方式，使得读者可以轻松理解信息的层次和结构。<br/> 6. 根据以上因素综合评估答案的语言组织，并给出一个 1 到 5 的分数，其中 5 表示语言组织非常好，而 1 表示语言组织非常差。</br></br>1. Read the answers and check for grammatical errors, poor word choice, or other significant mistakes.<br>2. Check that the answer is logical, conveys the information in a logical order, and is self-explanatory.<br>3. Determine if the answer is relevant to the question or topic and conveys a clear message.<br>4. Check that the answer is coherent and that appropriate transitions and switches are used to maintain coherence between sentences and paragraphs.<br>5. Check that the answer is clearly structured and organized in such a way that the reader can easily understand the hierarchy and structure of the information.<br>6. Evaluate the linguistic organization of the answer based on a combination of the above factors and give a score of 1 to 5, where 5 indicates very good linguistic organization and 1 indicates very poor linguistic organization. |
| 切题<br/>(Relevance) | 切题(1-5)：答案内容是否切题，不答非所问，并且严格遵照题目要求。</br></br>Relevance (1-5): whether the content of the answer is relevant to the topic, does not answer the wrong question, and strictly follows the requirements of the topic. | 1. 阅读题目，确定题目所问的问题是什么，以及需要回答哪些方面的问题。<br/> 2. 阅读答案，确认答案是否直接回答了题目所问的问题。<br/> 3. 检查答案是否严格遵照了题目的要求，包括答题方式、答题长度、答题格式等等。<br/> 4. 根据以上因素综合评估答案的切题程度，并给出一个 1 到 5 的分数，其中 5 表示答案非常切题，而 1 表示答案完全没有切题。</br></br>1. Read the question to determine what the question asks and what aspects of the question need to be answered.<br>2. Read the answers to make sure that they directly answer the question asked.<br>3. Check that the answer follows the requirements of the question, including the way it is answered, the length of the answer, the format of the answer, etc.<br>4. Evaluate how relevant the answer is based on the above factors and give a score of 1 to 5, where 5 means the answer is very relevant and 1 means the answer is not relevant at all. |
| 创意性<br/>(Creativity) | 创意性(1-5)：某些头脑风暴问题可能需要答案具有创意，提出新的思路。</br></br>Creativity (1-5): Some brainstorming questions may require answers that are creative and suggest new ideas. | 1. 仔细阅读所提供的头脑风暴问题，确保你理解问题的要点和背景。<br/> 2. 根据你的知识和经验，判断所提供的答案是否可行。如果答案不可行，则创意性评分可能会受到影响。<br/> 3. 考虑答案中是否包含新颖的想法或独特的思路。答案可能与已知的解决方案有所重叠，但仍然可以被认为是有创意的，只要它提供了新的角度或方法来解决问题。<br/> 4. 根据答案的创意性，给出一个 1 到 5 的评分。如果答案缺乏创意，则应给出一个较低的评分。如果答案具有创意并提供了新的思路，应给出一个较高的评分。</br></br>1. Read the provided brainstorming questions carefully to make sure you understand the gist and context of the questions.<br>2. Based on your knowledge and experience, determine if the answers provided are feasible. If the answer is not feasible, the creativity score may be affected.<br>3. Consider whether the answer contains novel ideas or unique thoughts. An answer may overlap with a known solution and still be considered creative, as long as it offers a new perspective or approach to the problem.<br>4. Give a score of 1 to 5 depending on the creativity of the answer. If the answer lacks creativity, a lower score should be given. If the answer is creative and provides a new idea, a higher score should be given. |
| 实用性<br/>(Practicality) | 实用性(1-5)：某些头脑风暴问题可能需要答案提出实用的建议或解决方法。</br></br>Practicality (1-5): Some brainstorming questions may require answers to suggest practical suggestions or solutions. | 1. 仔细阅读所提供的头脑风暴问题，确保你理解问题的要点和背景。<br/> 2. 根据你的知识和经验，判断所提供的答案是否可行。如果答案不可行，则实用性评分可能会受到影响。<br/> 3. 考虑答案中提出的建议或解决方法是否实用并可行。答案可能看起来很好，但如果无法实现或应用，则实用性评分可能会受到影响。<br/> 4. 根据答案的实用性，给出一个 1 到 5 的评分。如果答案缺乏实用性，则应给出一个较低的评分。如果答案提出了实用的建议或解决方法，并且可以很好地解决问题，则应给出一个较高的评分。</br></br>1. Read the provided brainstorming questions carefully to make sure you understand the gist and context of the questions.<br>2. Based on your knowledge and experience, determine if the answers provided are feasible. If the answer is not feasible, the practicality score may be affected.<br>3. Consider whether the suggestions or solutions presented in the answer are practical and workable. The answer may look good, but if it cannot be implemented or applied, the practicality score may be affected.<br>4. Give a score of 1 to 5 depending on the practicality of the answer. If the answer lacks practicality, a lower score should be given. If the answer makes a practical suggestion or solution and solves the problem well, a higher score should be given. |
| 正确性<br/>(Correctness) | 正确性(1-5)：正确性(1-5)：答案是否正确。</br></br> Correctness (1-5): whether the answer is correct or not. | 1. 仔细阅读题目，尝试自己回答该问题。<br/>2. 检查答案的准确性。您可以使用已知的事实或研究来验证答案是否正确。如果答案是正确的，则可以将正确性得分为 5 分。如果答案是部分正确的，则可以给予适当的得分，例如 2 分、3 分或 4 分。如果答案完全不正确，则只得 1 分。<br/><br/>1. Read the question carefully and try to answer the question yourself. <br/>2. Check the correctness of the answer. You can use known facts or research to verify that the answer is correct. If the answer is correct, you can give a score of 5 for correctness. If the answer is partially correct, an appropriate score, such as 2, 3, or 4, may be given. If the answer is completely incorrect, only 1 point is awarded. |
| 自然<br/>(Naturalness) | 自然(1-5)：答案是否自然，并且符合问题给定的身份。</br></br>Naturalness (1-5): whether the answer is natural and fits the identity given by the question. | 1. 阅读题目，确定题目提供的身份信息。<br/> 2. 检查答案内容是否符合题目给定的身份。<br/> 3. 根据以上因素，对该回答的自然性进行打分，分数从 1 到 5，其中 1 表示不自然，5 表示非常自然，并符合问题给定的身份。</br></br>1. Read the question and determine the identity information provided in the question.<br>2. Check whether the content of the answer matches the identity given in the question.<br>3. Based on the above factors, score the naturalness of the response on a scale from 1 to 5, where 1 means unnatural and 5 means very natural and in accordance with the identity given in the question. |
| 参与感<br/>(Engagingness) | 参与感(1-5)：答案是否对前面的对话内容做出了恰当的反应，是否理解对话的语境和背景。</br></br>Engagingness (1-5): whether the answer responds appropriately to the content of the preceding conversation and whether it understands the context and background of the conversation. | 1. 阅读题目，确定对话的语境和背景。<br/> 2. 检查答案是否充分理解对话的语境和背景，能否自然地融入到对话中而不显得突兀。<br/> 3. 根据以上因素，对该回答的参与感进行打分，分数从 1 到 5，其中 1 表示没有参与感，5 表示非常有参与感，并且恰当地理解了对话的语境和背景。</br></br>1. Read the questions to determine the context and background of the dialogue.<br>2. Check that the answer fully understands the context and background of the conversation and that it fits naturally into the conversation without seeming abrupt.<br>3. Based on the above factors, rate the response's engagement on a scale from 1 to 5, where 1 means not engaged and 5 means very engaged and appropriately understands the context and background of the conversation. |
| 合理性<br/>(Reasonableness) | 合理性(1-5)：答案是否能够与前面的对话内容形成逻辑上的衔接，是否符合常理，能否在这个上下文中合理存在。</br></br>Reasonableness (1-5): Whether the answer can form a logical connection with the content of the previous dialogue, whether it is consistent with common sense, and whether it can reasonably exist in this context. | 1. 阅读题目，确定对话的主题以及问题期望的回答方向。<br/> 2. 判断答案是否能够与前面的对话内容形成逻辑上的衔接，是否符合常理，能否在这个上下文中合理存在。<br/> 3. 根据以上因素，对该回答的合理性进行打分，分数从 1 到 5，其中 1 表示不合理，5 表示非常合理，并且能够与前面的对话内容形成逻辑上的衔接，并符合常理。</br></br>1. Read the question and determine the topic of the conversation and the direction the question expects the answer to go.<br>2. Determine whether the answer can be logically connected to the preceding conversation, whether it makes common sense, and whether it can reasonably exist in this context.<br>3. Based on the above factors, rate the reasonableness of the answer on a scale from 1 to 5, where 1 means unreasonable and 5 means very reasonable and able to form a logical connection with the preceding dialogue content and consistent with common sense. |
| 多样性<br/>(Diversity) | 多样性(1-5)：答案使用语言是否优美，具有有一定的创造性和想象力。然而，回答也应该保持合理和适度，不要过于夸张或离题。</br></br>Diversity (1-5): Whether the answers use beautiful language and have some creativity and imagination. However, answers should also be kept reasonable and moderate, not overly exaggerated or off-topic. | 1. 仔细阅读整个回答，确保完全理解回答所表达的内容和主题。<br/> 2. 在阅读回答的同时，注意语言的质量，例如措辞是否正确，语言是否生动等。<br/> 3. 检查回答的创造性和想象力，看看回答是否能够吸引人阅读下去。<br/> 4. 检查回答的合理性和适度，看看回答是否夸张或离题。5. 将多样性的评分打分在 1 到 5 之间，5 分表示回答的质量很好，能够吸引人阅读，1 分表示回答的内容生硬或者有离题的问题。</br></br>1. Read the entire response carefully to ensure that you fully understand the content and theme expressed in the response.<br>2. While reading the response, pay attention to the quality of the language, such as whether the wording is correct and the language is vivid.<br>3. Check the creativity and imagination of the response to see if the response is engaging to read on.<br>4. Check the reasonableness and appropriateness of the responses to see if the responses are exaggerated or off-topic.<br>5. Rate the diversity on a scale of 1 to 5, with a 5 indicating a good quality response that is engaging to read and a 1 indicating a raw response or a question that is off-topic. |
| 保真度<br/>(Fidelity) | 保真度(1-5)：答案是否能够严格遵守角色的设定回答给定的请求。</br></br>Fidelity (1-5): whether the answer is able to answer the given request in strict compliance with the role setting. | 1. 仔细阅读问题，了解角色在问题中的设定和表现，包括职业、背景、观点、性格等方面。<br/> 阅读题目的请求，确认回答请求时需要注意的细节。<br/> 3. 对比提供的回答与该角色的设定，评估回答是否能够严格遵守角色的设定。<br/> 4. 结合以上评估结果给出保真度的评分，范围从 1 到 5 分，其中 1 分表示回答与角色设定完全不符，5 分表示回答完全符合角色设定且满足给定请求。</br></br>1. Read the question carefully to understand how the character is set up and represented in the question, including aspects such as occupation, background, point of view, and personality.<br>2. Read the question's request and confirm the details that need to be taken into account when answering the request.<br>3. Compare the provided answer with the setting of the role and assess whether the answer can strictly adhere to the setting of the role.<br>4. Combine the results of the above assessment to give a fidelity score ranging from 1 to 5, where a score of 1 means that the response does not match the persona at all, and a score of 5 means that the response fully complies with the persona and satisfies the given request. |
| 简明扼要<br/>(Conciseness) | 简明扼要(1-5)：答案是否简明扼要，没有冗余内容。</br></br>Conciseness (1-5): answers should be concise and without redundant content. | 1. 阅读题目，提取出材料的重点。<br/> 2. 阅读该总结，并注意其中的主要观点和信息。<br/> 3. 评估总结的长度。一个简明扼要的总结通常应该在几句话或几段文字内传达关键信息，而不是冗长的段落或文章。<br/> 4. 检查总结是否包含与主要观点无关的信息或冗余信息。<br/> 5. 确定总结涵盖了材料中的关键信息，并且没有忽略任何重要细节。<br/> 6. 给总结打出 1-5 的分数，其中 5 表示总结简明扼要，没有冗余内容，而 1 表示总结冗长或包含不必要的信息，难以理解或记忆。根据您的判断，打出适当的得分。</br></br>1. Read the title and extract the main points of the material.<br>2. Read the summary and note the main ideas and messages in it.<br>3. Assess the length of the summary. A concise summary should usually convey key information within a few sentences or paragraphs, rather than lengthy paragraphs or essays.<br>4. Check that the summary does not contain information that is not relevant to the main ideas or that is redundant.<br>5. Make sure that the summary covers the key information in the material and that no important details have been omitted.<br>6. Rate the summary on a scale of 1-5, where 5 means the summary is concise and free of redundancy, and 1 means the summary is lengthy or contains unnecessary information that is difficult to understand or remember. Based on your judgment, assign the appropriate score. |
GPT models evaluate the quality of model predictions based on the given prompt words and gives a score between 1-5.
> **NOTE 1:** Even for the same metric, the details of its prompt words and CoT(Chain-of-Thought) can differ based on which category you want to evaluate. For example, prompt words for metric `correctness` showed here is "Whether the answer is correct or not."(this is for category `classification`), but for category `extraction`, prompt words can be "Answers should extract the required information accurately and should not contain any incorrect or misleading information." You can find all the prompt words and CoT(Chain-of-Thought) in `prompt/evaluation_prompt`.
> **NOTE 1:** Even for the same metric, the details of its prompt words and CoT(Chain-of-Thought) can differ based on which category you want to evaluate. For example, prompt words for metric `correctness` showed here is "Whether the answer is correct or not."(this is for category `classification`), but for category `extraction`, prompt words can be "Answers should extract the required information accurately and should not contain any incorrect or misleading information." You can find all the prompt words and CoT(Chain-of-Thought) in `prompt/evaluation_prompt`.
> **NOTE 2:** To add customized metrics, you can refer to [FAQ](#faq).
@ -86,19 +84,19 @@ GPT models evaluate the quality of model predictions based on the given prompt w
Automated metrics evaluate the capability of a model by comparing model predictions with reference answers.
There are two ways to obtain reference answers:
* For instruction coming from human-designed problems, the reference answers are generated by GPT-3.5, such as roleplay, chat.
* For instruction related with classic NLP problems, the reference answers are collected from open-sourced dataset with target answers, such as classification, extraction, summarization.
- For instruction coming from human-designed problems, the reference answers are generated by GPT-3.5, such as roleplay, chat.
- For instruction related with classic NLP problems, the reference answers are collected from open-sourced dataset with target answers, such as classification, extraction, summarization.
There are 6 types of automatic evaluation metrics listed in the table below:
| Automatic Evaluation Metric | Description |
| :---------------------------------: | :----------------------------------------------------------- |
| BLEU-n | Measure the accuracy between prediction and reference.<br/> BLEU-1 (Unigram) evaluates accuracy in word level.<br/> BLEU-n (n-gram) evaluate the fluency in sentence level. |
| Automatic Evaluation Metric | Description |
| :---------------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| BLEU-n | Measure the accuracy between prediction and reference.<br/> BLEU-1 (Unigram) evaluates accuracy in word level.<br/> BLEU-n (n-gram) evaluate the fluency in sentence level. |
| ROUGE | ROUGE-N measures the number of matching n-grams between prediction and reference. <br/> ROUGE-L measures the number of matching longest common subsequence (LCS) between prediction and reference. |
| Distinct | Measure the diversity of generation text by counting the unique n-grams. |
| BERTScore | Measure the semantic similarity between tokens of predictions and references with BERT. |
| Precision<br/> Recall<br/> F1 Score | Measure the number of overlaps between prediction and reference (design for classification and extraction categories). |
| CHRF | Measure the similarity of character n-grams between prediction and reference. |
| Distinct | Measure the diversity of generation text by counting the unique n-grams. |
| BERTScore | Measure the semantic similarity between tokens of predictions and references with BERT. |
| Precision<br/> Recall<br/> F1 Score | Measure the number of overlaps between prediction and reference (design for classification and extraction categories). |
| CHRF | Measure the similarity of character n-grams between prediction and reference. |
#### UniEval Evaluation
@ -106,17 +104,17 @@ UniEval converts all evaluation tasks of different dimensions(metrics) into Bool
In our evaluation pipeline, two pre-trained UniEval evaluators are used. One is [unieval-sum](https://huggingface.co/MingZhong/unieval-sum) and the other is [unieval-dialog](https://huggingface.co/MingZhong/unieval-dialog). The two models can be used for the 3 tasks, `summarization`, `dialogue` and `data2text`. Each task has different evaluation dimensions.
| UniEval Model | Task | Dimension(Metric) |
| :------------: | :----------------- | :--- |
| unieval-sum | summarization | coherence: whether the summary is coherent<br/>consistency: whether the claim is consistent with the given document<br/>fluency: whether the paragraph is fluent<br/>relevance: whether the summary is relevant to the reference |
| unieval-sum | data2text | naturalness: whether the utterance is fluent<br/>informativeness: whether the utterance is informative according to the reference |
| unieval-dialog | dialogue | naturalness: whether the response is natural in the dialogue<br/>coherence: whether the response is coherent in the dialogue history<br/>understandability: whether the response is understandable in the dialogue |
| UniEval Model | Task | Dimension(Metric) |
| :------------: | :------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| unieval-sum | summarization | coherence: whether the summary is coherent<br/>consistency: whether the claim is consistent with the given document<br/>fluency: whether the paragraph is fluent<br/>relevance: whether the summary is relevant to the reference |
| unieval-sum | data2text | naturalness: whether the utterance is fluent<br/>informativeness: whether the utterance is informative according to the reference |
| unieval-dialog | dialogue | naturalness: whether the response is natural in the dialogue<br/>coherence: whether the response is coherent in the dialogue history<br/>understandability: whether the response is understandable in the dialogue |
> **NOTE 1:** Task "data2text" uses the same model as task "summarization".
> **NOTE 1:** Task "data2text" uses the same model as task "summarization".
> **NOTE 2:** In UniEval paper, the `unieval-sum` model demonstrates the best transfer ability and so you can evaluate your customized metric with this model. Details of adding customized metrics can be found in [FAQ](#faq).
> **NOTE 2:** In UniEval paper, the `unieval-sum` model demonstrates the best transfer ability and so you can evaluate your customized metric with this model. Details of adding customized metrics can be found in [FAQ](#faq).
> **NOTE 3:** We consider not including all metrics provided in UniEval in our pipeline because the data structure and content of the instructions we want to evaluate are not suitable for direct use of some UniEval metrics.
> **NOTE 3:** We consider not including all metrics provided in UniEval in our pipeline because the data structure and content of the instructions we want to evaluate are not suitable for direct use of some UniEval metrics.
## Evaluation Process
@ -127,12 +125,12 @@ In our evaluation pipeline, two pre-trained UniEval evaluators are used. One is
A JSON file contains one list. Each element in the list is a target answer / prediction record for one instruction / question.
An element should have the following fields:
* `category` (str, compulsory): The category of the instruction / question.
* `instruction` (str, compulsory): The instruction / question for the LLM.
* `input` (str, optional): The additional context of the instruction / question.
* `output` (str, optional): The sample output of the instruction (default: GPT-3.5).
* `target` (str, optional): The target answer for the instruction.
* `id` (int, compulsory): The ID of the instruction / question.
- `category` (str, compulsory): The category of the instruction / question.
- `instruction` (str, compulsory): The instruction / question for the LLM.
- `input` (str, optional): The additional context of the instruction / question.
- `output` (str, optional): The sample output of the instruction (default: GPT-3.5).
- `target` (str, optional): The target answer for the instruction.
- `id` (int, compulsory): The ID of the instruction / question.
If the `input` has a target answer, the `output` can be empty. Otherwise, we generate answers from GPT-3.5 as the `output`, and the `target` field is empty.
@ -140,22 +138,22 @@ Example:
```json
[
{
"category": "brainstorming",
"instruction": "请介绍一下人工智能的多个领域。",
"input": "",
"output": "{GPT-3.5 Answers}",
"target": "",
"id": 1
},
{
"category": "classification",
"instruction": "新闻标题：为什么电影《倩女幽魂》中燕赤霞一个道士却拿着金刚经？请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。",
"input": "",
"output": "",
"target": "{target answer}",
"id": 2
}
{
"category": "brainstorming",
"instruction": "请介绍一下人工智能的多个领域。",
"input": "",
"output": "{GPT-3.5 Answers}",
"target": "",
"id": 1
},
{
"category": "classification",
"instruction": "新闻标题：为什么电影《倩女幽魂》中燕赤霞一个道士却拿着金刚经？请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。",
"input": "",
"output": "",
"target": "{target answer}",
"id": 2
}
]
```
@ -165,33 +163,33 @@ A JSON file contains one list. Each element in the list is a model answer / pred
An element should have the following fields:
* `category` (str, compulsory): The category of the instruction / question.
* `instruction` (str, compulsory): The instruction / question for the LLM.
* `input` (str, optional): The additional context of the instruction / question.
* `output` (str, compulsory): The output from the LLM.
* `target` (str, optional): The target answer for the instruction.
* `id` (int, compulsory): The ID of the instruction / question.
- `category` (str, compulsory): The category of the instruction / question.
- `instruction` (str, compulsory): The instruction / question for the LLM.
- `input` (str, optional): The additional context of the instruction / question.
- `output` (str, compulsory): The output from the LLM.
- `target` (str, optional): The target answer for the instruction.
- `id` (int, compulsory): The ID of the instruction / question.
Example:
```json
[
{
"category": "brainstorming",
"instruction": "请介绍一下人工智能的多个领域。",
"input": "",
"output": "{Model Answers / Predictions}",
"target": "",
"id": 1
},
{
"category": "classification",
"instruction": "新闻标题：为什么电影《倩女幽魂》中燕赤霞一个道士却拿着金刚经？请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。",
"input": "",
"output": "{Model Answers / Predictions}",
"target": "{target answer}",
"id": 2
}
{
"category": "brainstorming",
"instruction": "请介绍一下人工智能的多个领域。",
"input": "",
"output": "{Model Answers / Predictions}",
"target": "",
"id": 1
},
{
"category": "classification",
"instruction": "新闻标题：为什么电影《倩女幽魂》中燕赤霞一个道士却拿着金刚经？请根据新闻标题判断新闻所属的分类，你需要从文化，娱乐，体育，财经，房产，教育，科技，旅游，游戏，军事这十类中选择一个答案。",
"input": "",
"output": "{Model Answers / Predictions}",
"target": "{target answer}",
"id": 2
}
]
```
@ -212,7 +210,7 @@ The following is the Chinese battle prompt. In the battle prompt, the question a
#### Evaluation Prompt
The following is an example of a Chinese GPT evaluation prompt. In an evaluation prompt, you should define your metrics in `metrics` and provide CoT(Chain-of-Thought) in `CoT`. You can find example evaluation prompt files for Chinese and English in `prompt/evaluation_prompt`.
The following is an example of a Chinese GPT evaluation prompt. In an evaluation prompt, you should define your metrics in `metrics` and provide CoT(Chain-of-Thought) in `CoT`. You can find example evaluation prompt files for Chinese and English in `prompt/evaluation_prompt`.
```json
{
@ -242,24 +240,32 @@ The following is an example of a Chinese config file. The configuration file can
```json
{
"language": "en",
"path_for_UniEval": {
"summarization": "path to unieval-sum model",
"dialogue": "path to unieval-dialog model",
"data2text": "path to unieval-sum model"
"language": "en",
"path_for_UniEval": {
"summarization": "path to unieval-sum model",
"dialogue": "path to unieval-dialog model",
"data2text": "path to unieval-sum model"
},
"category": {
"brainstorming": {
"GPT": ["relevance", "creativity", "practicality", "reasonableness"],
"Metrics": ["Distinct"],
"UniEval": [
"summarization-fluency",
"data2text-naturalness",
"data2text-informativeness"
]
},
"category": {
"brainstorming": {
"GPT": ["relevance", "creativity", "practicality", "reasonableness"],
"Metrics": ["Distinct"],
"UniEval": ["summarization-fluency", "data2text-naturalness", "data2text-informativeness"]
},
"chat": {
"GPT": [ "relevance", "naturalness", "engagingness", "reasonableness"],
"Metrics": ["Distinct"],
"UniEval": ["dialogue-naturalness", "dialogue-coherence", "dialogue-understandability"]
}
"chat": {
"GPT": ["relevance", "naturalness", "engagingness", "reasonableness"],
"Metrics": ["Distinct"],
"UniEval": [
"dialogue-naturalness",
"dialogue-coherence",
"dialogue-understandability"
]
}
}
}
```
@ -293,7 +299,7 @@ You can create your config file based on available settings listed in following
| "summarization" | "fidelity" | | |
| | "conciseness" | | |
> **NOTE:** For categories which don't have standard answers such as `brainstorming`, you should avoid using automatic metrics such as `BLEU` and `ROUGE` which are based on similarity measures and you should use `Distinct` instead in your config file.
> **NOTE:** For categories which don't have standard answers such as `brainstorming`, you should avoid using automatic metrics such as `BLEU` and `ROUGE` which are based on similarity measures and you should use `Distinct` instead in your config file.
#### Evaluate
@ -346,8 +352,8 @@ For example, if you want to add a new metric `persuasiveness` into task `data2te
```python
if task == 'data2text':
if dimension == 'persuasiveness':
cur_input = 'question: Is this a persuasive utterence </s> utterance: ' + output[i]
if dimension == 'persuasiveness':
cur_input = 'question: Is this a persuasive utterence </s> utterance: ' + output[i]
```
</details>

260
applications/Chat/examples/README.md
View File

@ -17,7 +17,7 @@
- [Arg List](#arg-list-2)
- [Inference example - After Stage3](#inference-example---after-stage3)
- [Attention](#attention)
- [data](#data)
- [data](#data)
- [Support Model](#support-model)
- [GPT](#gpt)
- [BLOOM](#bloom)
@ -28,8 +28,8 @@
- [Reward model](#reward-model)
- [Critic model](#critic-model)
---
## Install requirements
```shell
@ -38,10 +38,11 @@ pip install -r requirements.txt
## Supervised datasets collection
We collected 104K bilingual dataset of Chinese and English, and you can find the datasets in this repo
[InstructionWild](https://github.com/XueFuzhao/InstructionWild).
We collected 104K bilingual datasets of Chinese and English, and you can find the datasets in this repo
[InstructionWild](https://github.com/XueFuzhao/InstructionWild) and in this [file](https://github.com/XueFuzhao/InstructionWild/blob/main/data/README.md).
Here is how we collected the data
The following pic shows how we collected the data.
<p align="center">
<img src="https://raw.githubusercontent.com/hpcaitech/public_assets/main/applications/chat/data-collect.png" width=500/>
</p>
@ -52,38 +53,40 @@ In order to further improve the model's ability to handle multi-turn conversatio
A sample of conversation dataset should have the following fields:
* `type` (str, optional): The type of the data sample.
* `language` (str, optional): The language of the data sample.
* `dataset` (str, optional): The dataset the data sample originates from.
* `conversations` (str, compulsory): Conversation content of the data sample.
* `id` (int, optional): The ID of the data sample.
- `type` (str, optional): The type of the data sample.
- `language` (str, optional): The language of the data sample.
- `dataset` (str, optional): The dataset the data sample originates from.
- `conversations` (str, compulsory): Conversation content of the data sample.
- `id` (int, optional): The ID of the data sample.
A simple example:
```json
{
"type": "instruction",
"language": "English",
"dataset": "Alpaca",
"conversations": [
{
"from": "human",
"value": "Give three tips for staying healthy."
},
{
"from": "gpt",
"value": "1.Eat a balanced diet and make sure to include plenty of fruits and vegetables. \n2. Exercise regularly to keep your body active and strong. \n3. Get enough sleep and maintain a consistent sleep schedule."
}
],
"id": 1
"type": "instruction",
"language": "English",
"dataset": "Alpaca",
"conversations": [
{
"from": "human",
"value": "Give three tips for staying healthy."
},
{
"from": "gpt",
"value": "1.Eat a balanced diet and make sure to include plenty of fruits and vegetables. \n2. Exercise regularly to keep your body active and strong. \n3. Get enough sleep and maintain a consistent sleep schedule."
}
],
"id": 1
}
```
> **NOTE:** Only key `conversations` is compulsary for training and other keys serve as metadata. The length of `conversations` varies.
> **NOTE:** Only key `conversations` is compulsary for training and other keys serve as metadata. The length of `conversations` varies.
You can run the `examples/generate_conversation_dataset.py` to generate a conversation dataset supported by ColossalChat.
You can use the following cmd to generate conversation dataset.
```
```bash
python generate_conversation_dataset.py \
--dataset "All"
--save_path "/path/to/dataset"
@ -97,12 +100,12 @@ Stage1 is supervised instructs fine-tuning, which uses the datasets mentioned ea
You can run the `examples/train_sft.sh` to start a supervised instructs fine-tuning.
You can also use the following cmd to start a supervised instructs fine-tuning with your own settings.
```
```bash
torchrun --standalone --nproc_per_node=4 train_sft.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy colossalai_zero2 \
--log_interval 10 \
--save_path /path/to/Coati-7B \
--dataset /path/to/data.json \
--batch_size 4 \
@ -112,18 +115,33 @@ torchrun --standalone --nproc_per_node=4 train_sft.py \
--max_epochs 1 \
--grad_checkpoint
```
**Note**: the supervised dataset follows the following format,
```json
[
{
"instruction": "Provide a list of the top 10 most popular mobile games in Asia",
"input": "",
"output": "The top 10 most popular mobile games in Asia are:\n1) PUBG Mobile\n2) Pokemon Go\n3) Candy Crush Saga\n4) Free Fire\n5) Clash of Clans\n6) Mario Kart Tour\n7) Arena of Valor\n8) Fantasy Westward Journey\n9) Subway Surfers\n10) ARK Survival Evolved",
"id": 0
},
...
]
```
### Arg List
- --strategy: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- --model: model type, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- --pretrain: pretrain model, type=str, default=None
- --max_datasets_size: the max size of dataset, type=int, default=None
- --save_path: path to save the model, type=str, default='output'
- --need_optim_ckpt: whether to save optim ckpt, type=bool, default=False
- --max_epochs: max epochs for training, type=int, default=3
- --batch_size: batch size while training, type=int, default=4
- --lora_rank: low-rank adaptation matrices rank, type=int, default=0
- --log_interval: how many steps to log, type=int, default=100
- --grad_checkpoint: enable gradient checkpointing, type=bool, default=False
- `--strategy`: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- `--model`: model type, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- `--pretrain`: pretrain model, type=str, default=None
- `--max_datasets_size`: the max size of dataset, type=int, default=None
- `--save_path`: path to save the model, type=str, default='output'
- `--need_optim_ckpt`: whether to save optim ckpt, type=bool, default=False
- `--max_epochs`: max epochs for training, type=int, default=3
- `--batch_size`: batch size while training, type=int, default=4
- `--lora_rank`: low-rank adaptation matrices rank, type=int, default=0
- `--grad_checkpoint`: enable gradient checkpointing, type=bool, default=False
## Stage2 - Training reward model
@ -133,7 +151,8 @@ We train a reward model in stage 2, which obtains corresponding scores by manual
You can run the `examples/train_rm.sh` to start a reward model training.
You can also use the following cmd to start training a reward model.
```
```bash
torchrun --standalone --nproc_per_node=4 train_reward_model.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
@ -141,16 +160,19 @@ torchrun --standalone --nproc_per_node=4 train_reward_model.py \
--loss_fn 'log_exp'\
--save_path 'rmstatic.pt' \
```
### Features and tricks in RM training
- We support [Anthropic/hh-rlhf](https://huggingface.co/datasets/Anthropic/hh-rlhf)and[rm-static](https://huggingface.co/datasets/Dahoas/rm-static) datasets.
- We support 2 kinds of loss_function named 'log_sig'(used by OpenAI) and 'log_exp'(used by Anthropic).
- We change the loss to valid_acc and pair_dist to monitor progress during training.
- We support 2 kinds of loss function named `log_sig`(used by OpenAI) and `log_exp`(used by Anthropic).
- We change the loss to `valid_acc` and `pair_dist` to monitor progress during training.
- We add special token to the end of the sequence to get better result.
- We use cosine-reducing lr-scheduler for RM training.
- We set value_head as 1 liner layer and initialize the weight of value_head using N(0，1/(d_model + 1)) distribution.
- We train a Bloom-560m reward model for 1 epoch and find the test acc of the model achieve the performance mentions in [Anthropics paper](https://arxiv.org/abs/2204.05862).
### Experiment result
Model performance in [Anthropics paper](https://arxiv.org/abs/2204.05862):
<div align=middle> <img width="512" alt="image" src="https://user-images.githubusercontent.com/70618399/225263321-8d64c3a8-6877-4cc8-9b61-0e1c52d3d94f.png">
@ -162,20 +184,20 @@ Model performance in [Anthropics paper](https://arxiv.org/abs/2204.05862):
<div align=left>We also train the reward model based on LLaMA-7B, which reaches the ACC of 72.06% after 1 epoch, performing almost the same as Anthropic's best RM.
### Arg List
- --strategy: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- --model: model type, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- --pretrain: pretrain model, type=str, default=None
- --model_path: the path of rm model(if continue to train), type=str, default=None
- --save_path: path to save the model, type=str, default='output'
- --need_optim_ckpt: whether to save optim ckpt, type=bool, default=False
- --max_epochs: max epochs for training, type=int, default=3
- --dataset: dataset name, type=str, choices=['Anthropic/hh-rlhf', 'Dahoas/rm-static']
- --subset: subset of the dataset, type=str, default=None
- --batch_size: batch size while training, type=int, default=4
- --lora_rank: low-rank adaptation matrices rank, type=int, default=0
- --loss_func: which kind of loss function, choices=['log_sig', 'log_exp']
- --max_len: max sentence length for generation, type=int, default=512
- --test: whether is only testing, if it's true, the dataset will be small
- `--strategy`: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- `--model`: model type, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- `--pretrain`: pretrain model, type=str, default=None
- `--model_path`: the path of rm model(if continue to train), type=str, default=None
- `--save_path`: path to save the model, type=str, default='output'
- `--need_optim_ckpt`: whether to save optim ckpt, type=bool, default=False
- `--max_epochs`: max epochs for training, type=int, default=3
- `--dataset`: dataset name, type=str, choices=['Anthropic/hh-rlhf', 'Dahoas/rm-static']
- `--subset`: subset of the dataset, type=str, default=None
- `--batch_size`: batch size while training, type=int, default=4
- `--lora_rank`: low-rank adaptation matrices rank, type=int, default=0
- `--loss_func`: which kind of loss function, choices=['log_sig', 'log_exp']
- `--max_len`: max sentence length for generation, type=int, default=512
## Stage3 - Training model using prompts with RL
@ -186,53 +208,89 @@ Stage3 uses reinforcement learning algorithm, which is the most complex part of
</p>
You can run the `examples/train_prompts.sh` to start PPO training.
You can also use the cmd following to start PPO training.
[[Stage3 tutorial video]](https://www.youtube.com/watch?v=Z8wwSHxPL9g)
```
```bash
torchrun --standalone --nproc_per_node=4 train_prompts.py \
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy colossalai_zero2 \
--prompt_dataset /path/to/your/prompt_dataset \
--pretrain_dataset /path/to/your/pretrain_dataset \
--rm_pretrain /your/pretrain/rm/definition \
--rm_path /your/rm/model/path
--pretrain "/path/to/LLaMa-7B/" \
--model 'llama' \
--strategy colossalai_zero2 \
--prompt_dataset /path/to/your/prompt_dataset \
--pretrain_dataset /path/to/your/pretrain_dataset \
--rm_pretrain /your/pretrain/rm/definition \
--rm_path /your/rm/model/path
```
Prompt dataset: the instruction dataset mentioned in the above figure which includes the instructions, e.g. you can use the [script](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples/generate_prompt_dataset.py) which samples `instinwild_en.json` or `instinwild_ch.json` in [InstructionWild](https://github.com/XueFuzhao/InstructionWild/tree/main/data#instructwild-data) to generate the prompt dataset.
Pretrain dataset: the pretrain dataset including the instruction and corresponding response, e.g. you can use the [InstructWild Data](https://github.com/XueFuzhao/InstructionWild/tree/main/data) in stage 1 supervised instructs tuning.
**Note**: the required datasets follow the following format,
- `pretrain dataset`
```json
[
{
"instruction": "Provide a list of the top 10 most popular mobile games in Asia",
"input": "",
"output": "The top 10 most popular mobile games in Asia are:\n1) PUBG Mobile\n2) Pokemon Go\n3) Candy Crush Saga\n4) Free Fire\n5) Clash of Clans\n6) Mario Kart Tour\n7) Arena of Valor\n8) Fantasy Westward Journey\n9) Subway Surfers\n10) ARK Survival Evolved",
"id": 0
},
...
]
```
- `prompt dataset`
```json
[
{
"instruction": "Edit this paragraph to make it more concise: \"Yesterday, I went to the store and bought some things. Then, I came home and put them away. After that, I went for a walk and met some friends.\"",
"id": 0
},
{
"instruction": "Write a descriptive paragraph about a memorable vacation you went on",
"id": 1
},
...
]
```
### Arg List
- --strategy: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- --model: model type of actor, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- --pretrain: pretrain model, type=str, default=None
- --rm_model: reward model type, type=str, choices=['gpt2', 'bloom', 'opt', 'llama'], default=None
- --rm_pretrain: pretrain model for reward model, type=str, default=None
- --rm_path: the path of rm model, type=str, default=None
- --save_path: path to save the model, type=str, default='output'
- --prompt_dataset: path of the prompt dataset, type=str, default=None
- --pretrain_dataset: path of the ptx dataset, type=str, default=None
- --need_optim_ckpt: whether to save optim ckpt, type=bool, default=False
- --num_episodes: num of episodes for training, type=int, default=10
- --num_update_steps: number of steps to update policy per episode, type=int
- --num_collect_steps: number of steps to collect experience per episode, type=int
- --train_batch_size: batch size while training, type=int, default=8
- --ptx_batch_size: batch size to compute ptx loss, type=int, default=1
- --experience_batch_size: batch size to make experience, type=int, default=8
- --lora_rank: low-rank adaptation matrices rank, type=int, default=0
- --kl_coef: kl_coef using for computing reward, type=float, default=0.1
- --ptx_coef: ptx_coef using for computing policy loss, type=float, default=0.9
- `--strategy`: the strategy using for training, choices=['ddp', 'colossalai_gemini', 'colossalai_zero2'], default='colossalai_zero2'
- `--model`: model type of actor, choices=['gpt2', 'bloom', 'opt', 'llama'], default='bloom'
- `--pretrain`: pretrain model, type=str, default=None
- `--rm_model`: reward model type, type=str, choices=['gpt2', 'bloom', 'opt', 'llama'], default=None
- `--rm_pretrain`: pretrain model for reward model, type=str, default=None
- `--rm_path`: the path of rm model, type=str, default=None
- `--save_path`: path to save the model, type=str, default='output'
- `--prompt_dataset`: path of the prompt dataset, type=str, default=None
- `--pretrain_dataset`: path of the ptx dataset, type=str, default=None
- `--need_optim_ckpt`: whether to save optim ckpt, type=bool, default=False
- `--num_episodes`: num of episodes for training, type=int, default=10
- `--num_update_steps`: number of steps to update policy per episode, type=int
- `--num_collect_steps`: number of steps to collect experience per episode, type=int
- `--train_batch_size`: batch size while training, type=int, default=8
- `--ptx_batch_size`: batch size to compute ptx loss, type=int, default=1
- `--experience_batch_size`: batch size to make experience, type=int, default=8
- `--lora_rank`: low-rank adaptation matrices rank, type=int, default=0
- `--kl_coef`: kl_coef using for computing reward, type=float, default=0.1
- `--ptx_coef`: ptx_coef using for computing policy loss, type=float, default=0.9
## Inference example - After Stage3
We support different inference options, including int8 and int4 quantization.
For details, see [`inference/`](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/inference).
## Attention
The examples are demos for the whole training process.You need to change the hyper-parameters to reach great performance.
#### data
- [x] [rm-static](https://huggingface.co/datasets/Dahoas/rm-static)
- [x] [hh-rlhf](https://huggingface.co/datasets/Anthropic/hh-rlhf)
- [ ] [openai/summarize_from_feedback](https://huggingface.co/datasets/openai/summarize_from_feedback)
@ -242,14 +300,16 @@ The examples are demos for the whole training process.You need to change the hyp
## Support Model
### GPT
- [x] GPT2-S (s)
- [x] GPT2-M (m)
- [x] GPT2-L (l)
- [x] GPT2-XL (xl)
- [x] GPT2-4B (4b)
- [ ] GPT2-6B (6b)
- [x] GPT2-S (s)
- [x] GPT2-M (m)
- [x] GPT2-L (l)
- [x] GPT2-XL (xl)
- [x] GPT2-4B (4b)
- [ ] GPT2-6B (6b)
### BLOOM
- [x] [BLOOM-560m](https://huggingface.co/bigscience/bloom-560m)
- [x] [BLOOM-1b1](https://huggingface.co/bigscience/bloom-1b1)
- [x] [BLOOM-3b](https://huggingface.co/bigscience/bloom-3b)
@ -257,6 +317,7 @@ The examples are demos for the whole training process.You need to change the hyp
- [ ] [BLOOM-175b](https://huggingface.co/bigscience/bloom)
### OPT
- [x] [OPT-125M](https://huggingface.co/facebook/opt-125m)
- [x] [OPT-350M](https://huggingface.co/facebook/opt-350m)
- [x] [OPT-1.3B](https://huggingface.co/facebook/opt-1.3b)
@ -266,10 +327,11 @@ The examples are demos for the whole training process.You need to change the hyp
- [ ] [OPT-30B](https://huggingface.co/facebook/opt-30b)
### [LLaMA](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md)
- [x] LLaMA-7B
- [x] LLaMA-13B
- [ ] LLaMA-33B
- [ ] LLaMA-65B
- [x] LLaMA-7B
- [x] LLaMA-13B
- [ ] LLaMA-33B
- [ ] LLaMA-65B
## Add your own models
@ -282,12 +344,12 @@ if it is supported in huggingface [transformers](https://github.com/huggingface/
r you can build your own model by yourself.
### Actor model
```
```python
from ..base import Actor
from transformers.models.coati import CoatiModel
class CoatiActor(Actor):
def __init__(self,
pretrained: Optional[str] = None,
checkpoint: bool = False,
@ -302,7 +364,8 @@ class CoatiActor(Actor):
```
### Reward model
```
```python
from ..base import RewardModel
from transformers.models.coati import CoatiModel
@ -325,12 +388,11 @@ class CoatiRM(RewardModel):
### Critic model
```
```python
from ..base import Critic
from transformers.models.coati import CoatiModel
class CoatiCritic(Critic):
def __init__(self,
pretrained: Optional[str] = None,
checkpoint: bool = False,

15
applications/Chat/examples/community/README.md
View File

@ -1,5 +1,9 @@
:warning: **This content may be outdated since the major update of Colossal Chat. We will update this content soon.**
# Community Examples
---
We are thrilled to announce the latest updates to ColossalChat, an open-source solution for cloning ChatGPT with a complete RLHF (Reinforcement Learning with Human Feedback) pipeline.
As Colossal-AI undergoes major updates, we are actively maintaining ColossalChat to stay aligned with the project's progress. With the introduction of Community-driven example, we aim to create a collaborative platform for developers to contribute exotic features built on top of ColossalChat.
@ -14,11 +18,12 @@ For more information about community pipelines, please have a look at this [issu
Community examples consist of both inference and training examples that have been added by the community. Please have a look at the following table to get an overview of all community examples. Click on the Code Example to get a copy-and-paste ready code example that you can try out. If a community doesn't work as expected, please open an issue and ping the author on it.
| Example | Description | Code Example | Colab | Author |
|:---------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------:|
| Peft | Adding Peft support for SFT and Prompts model training | [Huggingface Peft](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples/community/peft) | - | [YY Lin](https://github.com/yynil) |
| Train prompts on Ray | A Ray based implementation of Train prompts example | [Training On Ray](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples/community/ray) | - | [MisterLin1995](https://github.com/MisterLin1995) |
|...|...|...|...|...|
| Example | Description | Code Example | Colab | Author |
| :------------------- | :----------------------------------------------------- | :-------------------------------------------------------------------------------------------------------------- | :---- | ------------------------------------------------: |
| Peft | Adding Peft support for SFT and Prompts model training | [Huggingface Peft](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples/community/peft) | - | [YY Lin](https://github.com/yynil) |
| Train prompts on Ray | A Ray based implementation of Train prompts example | [Training On Ray](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples/community/ray) | - | [MisterLin1995](https://github.com/MisterLin1995) |
| ... | ... | ... | ... | ... |
### How to get involved
To join our community-driven initiative, please visit the [ColossalChat GitHub repository](https://github.com/hpcaitech/ColossalAI/tree/main/applications/Chat/examples), review the provided information, and explore the codebase. To contribute, create a new issue outlining your proposed feature or enhancement, and our team will review and provide feedback. We look forward to collaborating with you on this exciting project!

6
applications/Chat/examples/community/peft/README.md
View File

@ -1,3 +1,5 @@
:warning: **This content may be outdated since the major update of Colossal Chat. We will update this content soon.**
# Add Peft support for SFT and Prompts model training
The original implementation just adopts the loralib and merges the layers into the final model. The huggingface peft is a better lora model implementation and can be easily training and distributed.
@ -5,7 +7,9 @@ The original implementation just adopts the loralib and merges the layers into t
Since reward model is relative small, I just keep it as original one. I suggest train full model to get the proper reward/critic model.
# Preliminary installation
Since the current pypi peft package(0.2) has some bugs, please install the peft package using source.
```
git clone https://github.com/huggingface/peft
cd peft
@ -13,6 +17,7 @@ pip install .
```
# Usage
For SFT training, just call train_peft_sft.py
Its arguments are almost identical to train_sft.py instead adding a new eval_dataset if you have a eval_dataset file. The data file is just a plain datafile, please check the format in the easy_dataset.py.
@ -21,4 +26,5 @@ For stage-3 rlhf training, call train_peft_prompts.py.
Its arguments are almost identical to train_prompts.py. The only difference is that I use text files to indicate the prompt and pretrained data file. The models are included in easy_models.py. Currently only bloom models are tested, but technically gpt2/opt/llama should be supported.
# Dataformat
Please refer the formats in test_sft.txt, test_prompts.txt, test_pretrained.txt.

14
applications/Chat/examples/community/ray/README.md
View File

@ -1,17 +1,31 @@
:warning: **This content may be outdated since the major update of Colossal Chat. We will update this content soon.**
# ColossalAI on Ray
## Abstract
This is an experimental effort to run ColossalAI Chat training on Ray
## How to use?
### 1. Setup Ray clusters
Please follow the official [Ray cluster setup instructions](https://docs.ray.io/en/latest/cluster/getting-started.html) to setup an cluster with GPU support. Record the cluster's api server endpoint, it should be something similar to http://your.head.node.addrees:8265
### 2. Clone repo
Clone this project:
```shell
git clone https://github.com/hpcaitech/ColossalAI.git
```
### 3. Submit the ray job
```shell
python applications/Chat/examples/community/ray/ray_job_script.py http://your.head.node.addrees:8265
```
### 4. View your job on the Ray Dashboard
Open your ray cluster dashboard http://your.head.node.addrees:8265 to view your submitted training job.

24
applications/Chat/inference/README.md
View File

@ -20,21 +20,21 @@ Tha data is from [LLaMA Int8 4bit ChatBot Guide v2](https://rentry.org/llama-tar
### 8-bit
| Model | Min GPU RAM | Recommended GPU RAM | Min RAM/Swap | Card examples |
| :---: | :---: | :---: | :---: | :---: |
| LLaMA-7B | 9.2GB | 10GB | 24GB | 3060 12GB, RTX 3080 10GB, RTX 3090 |
| LLaMA-13B | 16.3GB | 20GB | 32GB | RTX 3090 Ti, RTX 4090 |
| LLaMA-30B | 36GB | 40GB | 64GB | A6000 48GB, A100 40GB |
| LLaMA-65B | 74GB | 80GB | 128GB | A100 80GB |
| Model | Min GPU RAM | Recommended GPU RAM | Min RAM/Swap | Card examples |
| :-------: | :---------: | :-----------------: | :----------: | :--------------------------------: |
| LLaMA-7B | 9.2GB | 10GB | 24GB | 3060 12GB, RTX 3080 10GB, RTX 3090 |
| LLaMA-13B | 16.3GB | 20GB | 32GB | RTX 3090 Ti, RTX 4090 |
| LLaMA-30B | 36GB | 40GB | 64GB | A6000 48GB, A100 40GB |
| LLaMA-65B | 74GB | 80GB | 128GB | A100 80GB |
### 4-bit
| Model | Min GPU RAM | Recommended GPU RAM | Min RAM/Swap | Card examples |
| :---: | :---: | :---: | :---: | :---: |
| LLaMA-7B | 3.5GB | 6GB | 16GB | RTX 1660, 2060, AMD 5700xt, RTX 3050, 3060 |
| LLaMA-13B | 6.5GB | 10GB | 32GB | AMD 6900xt, RTX 2060 12GB, 3060 12GB, 3080, A2000 |
| LLaMA-30B | 15.8GB | 20GB | 64GB | RTX 3080 20GB, A4500, A5000, 3090, 4090, 6000, Tesla V100 |
| LLaMA-65B | 31.2GB | 40GB | 128GB | A100 40GB, 2x3090, 2x4090, A40, RTX A6000, 8000, Titan Ada |
| Model | Min GPU RAM | Recommended GPU RAM | Min RAM/Swap | Card examples |
| :-------: | :---------: | :-----------------: | :----------: | :--------------------------------------------------------: |
| LLaMA-7B | 3.5GB | 6GB | 16GB | RTX 1660, 2060, AMD 5700xt, RTX 3050, 3060 |
| LLaMA-13B | 6.5GB | 10GB | 32GB | AMD 6900xt, RTX 2060 12GB, 3060 12GB, 3080, A2000 |
| LLaMA-30B | 15.8GB | 20GB | 64GB | RTX 3080 20GB, A4500, A5000, 3090, 4090, 6000, Tesla V100 |
| LLaMA-65B | 31.2GB | 40GB | 128GB | A100 40GB, 2x3090, 2x4090, A40, RTX A6000, 8000, Titan Ada |
## General setup