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

remove vl 4bit test case

pull/710/head
zhulinJulia24 2024-02-26 16:09:57 +08:00 committed by GitHub
parent f304eaa56d
commit 39170d27ef
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 9 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
tests/test_hf_model.py
View File

@ -1,6 +1,6 @@
import pytest import pytest
import torch, auto_gptq import torch, auto_gptq
from auto_gptq.modeling import BaseGPTQForCausalLM #from auto_gptq.modeling import BaseGPTQForCausalLM
from PIL import Image from PIL import Image
from transformers import AutoModel, AutoModelForCausalLM, AutoTokenizer from transformers import AutoModel, AutoModelForCausalLM, AutoTokenizer
@ -123,16 +123,11 @@ class TestMMModel:
trust_remote_code=True) trust_remote_code=True)
# Set `torch_dtype=torch.float16` to load model in float16, otherwise # Set `torch_dtype=torch.float16` to load model in float16, otherwise
# it will be loaded as float32 and might cause OOM Error. # it will be loaded as float32 and might cause OOM Error.
if '4bit' in model_name:
model = InternLMXComposer2QForCausalLM.from_quantized( model = AutoModelForCausalLM.from_pretrained(
model_name, trust_remote_code=True, device='cuda:0').eval() model_name, torch_dtype=torch.float16,
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True).cuda()
trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained(model_name,
else:
model = AutoModelForCausalLM.from_pretrained(
model_name, torch_dtype=torch.float16,
trust_remote_code=True).cuda()
tokenizer = AutoTokenizer.from_pretrained(model_name,
trust_remote_code=True) trust_remote_code=True)
model = model.eval() model = model.eval()
@ -178,7 +173,6 @@ class TestMMVlModel:
'model_name', 'model_name',
[ [
'internlm/internlm-xcomposer2-vl-7b', 'internlm/internlm-xcomposer2-vl-7b',
'internlm/internlm-xcomposer2-vl-7b-4bit'
], ],
) )
def test_demo_default(self, model_name): def test_demo_default(self, model_name):
@ -188,15 +182,9 @@ class TestMMVlModel:
torch.set_grad_enabled(False) torch.set_grad_enabled(False)
# init model and tokenizer # init model and tokenizer
if '4bit' in model_name: model = AutoModel.from_pretrained(
model = InternLMXComposer2QForCausalLM.from_quantized( model_name, trust_remote_code=True).cuda().eval()
model_name, trust_remote_code=True, device='cuda:0').eval() tokenizer = AutoTokenizer.from_pretrained(model_name,
tokenizer = AutoTokenizer.from_pretrained(model_name,
trust_remote_code=True)
else:
model = AutoModel.from_pretrained(
model_name, trust_remote_code=True).cuda().eval()
tokenizer = AutoTokenizer.from_pretrained(model_name,
trust_remote_code=True) trust_remote_code=True)
query = '<ImageHere>Please describe this image in detail.' query = '<ImageHere>Please describe this image in detail.'
@ -212,19 +200,3 @@ class TestMMVlModel:
assert 'Oscar Wilde' in response assert 'Oscar Wilde' in response
assert 'Live life with no excuses, travel with no regret' in response assert 'Live life with no excuses, travel with no regret' in response
class InternLMXComposer2QForCausalLM(BaseGPTQForCausalLM):
layers_block_name = 'model.layers'
outside_layer_modules = [
'vit',
'vision_proj',
'model.tok_embeddings',
'model.norm',
'output',
]
inside_layer_modules = [
['attention.wqkv.linear'],
['attention.wo.linear'],
['feed_forward.w1.linear', 'feed_forward.w3.linear'],
['feed_forward.w2.linear'],
]