[coati] fix inference output (#3285)

* [coati] fix inference requirements

* [coati] add output postprocess

* [coati] update inference readme

* [coati] fix inference requirements

applications/Chat/inference/README.md

@@ -36,6 +36,12 @@ Tha data is from [LLaMA Int8 4bit ChatBot Guide v2](https://rentry.org/llama-tar
 | 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
+
+```shell
+pip install -r requirements.txt
+```
+
 ## 8-bit setup
 
 8-bit quantization is originally supported by the latest [transformers](https://github.com/huggingface/transformers). Please install it from source.

applications/Chat/inference/requirements.txt

@@ -1,5 +1,5 @@
 fastapi
-locustio
+locust
 numpy
 pydantic
 safetensors
@@ -8,3 +8,5 @@ sse_starlette
 torch
 uvicorn
 git+https://github.com/huggingface/transformers
+accelerate
+bitsandbytes

applications/Chat/inference/server.py

@@ -17,7 +17,7 @@ from transformers import AutoTokenizer, GenerationConfig, LlamaForCausalLM
 from utils import ChatPromptProcessor, Dialogue, LockedIterator, sample_streamingly, update_model_kwargs_fn
 
 CONTEXT = 'Below is an instruction that describes a task. Write a response that appropriately completes the request. Do not generate new instructions.'
-MAX_LEN = 2048
+MAX_LEN = 512
 running_lock = Lock()
 
 
@@ -116,7 +116,7 @@ def generate_no_stream(data: GenerationTaskReq, request: Request):
     prompt_len = inputs['input_ids'].size(1)
     response = output[0, prompt_len:]
     out_string = tokenizer.decode(response, skip_special_tokens=True)
-    return out_string.lstrip()
+    return prompt_processor.postprocess_output(out_string)
 
 
 if __name__ == '__main__':

applications/Chat/inference/utils.py

@@ -1,3 +1,4 @@
+import re
 from threading import Lock
 from typing import Any, Callable, Generator, List, Optional
 
@@ -118,6 +119,9 @@ def _format_dialogue(instruction: str, response: str = ''):
     return f'\n\n### Instruction:\n{instruction}\n\n### Response:\n{response}'
 
 
+STOP_PAT = re.compile(r'(###|instruction:).*', flags=(re.I | re.S))
+
+
 class ChatPromptProcessor:
 
     def __init__(self, tokenizer, context: str, max_len: int = 2048):
@@ -164,6 +168,10 @@ class ChatPromptProcessor:
         prompt += ''.join(rows) + _format_dialogue(last_dialogue.instruction)
         return prompt
 
+    def postprocess_output(self, output: str) -> str:
+        output = STOP_PAT.sub('', output)
+        return output.strip()
+
 
 class LockedIterator: