add more models into daily testcases (InternLM#717)

Co-authored-by: zhulin1 <zhulin1@pjlab.org.cn>

.github/workflows/daily_tests.yaml

@@ -27,7 +27,7 @@ jobs:
         conda create -n internlm-model-latest --clone ${CONDA_BASE_ENV}
         source activate internlm-model-latest
         pip install transformers==${{ matrix.transformers-version }}
-        pip install sentencepiece auto-gptq
+        pip install sentencepiece auto-gptq==0.6.0 lmdeploy[all]
         srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=${GITHUB_RUN_ID}-${GITHUB_JOB} --gpus-per-task=2 pytest -s -v --color=yes ./tests/test_hf_model.py
         conda deactivate
     - name: load_latest_hf_model
@@ -36,7 +36,7 @@ jobs:
         conda create -n internlm-model-latest --clone ${CONDA_BASE_ENV}
         source activate internlm-model-latest
         pip install transformers
-        pip install sentencepiece auto-gptq
+        pip install sentencepiece auto-gptq==0.6.0 lmdeploy[all]
         srun -p ${SLURM_PARTITION} --kill-on-bad-exit=1 --job-name=${GITHUB_RUN_ID}-${GITHUB_JOB} --gpus-per-task=2 pytest -s -v --color=yes ./tests/test_hf_model.py
         conda deactivate
     - name: remove_env

tests/test_hf_model.py

@@ -1,5 +1,7 @@
 import pytest
 import torch
+from auto_gptq.modeling import BaseGPTQForCausalLM
+from lmdeploy import TurbomindEngineConfig, pipeline
 from PIL import Image
 from transformers import AutoModel, AutoModelForCausalLM, AutoTokenizer
 
@@ -9,6 +11,8 @@
 def assert_model(response):
     assert len(response) != 0
     assert 'UNUSED_TOKEN' not in response
+    assert 'Mynameis' not in response
+    assert 'Iama' not in response
 
 
 class TestChat:
@@ -18,6 +22,7 @@ class TestChat:
         'model_name',
         [
             'internlm/internlm2-chat-7b', 'internlm/internlm2-chat-7b-sft',
+            'internlm/internlm2-chat-20b', 'internlm/internlm2-chat-20b-sft',
             'internlm/internlm2-chat-1_8b', 'internlm/internlm2-chat-1_8b-sft'
         ],
     )
@@ -55,13 +60,31 @@ def test_demo_default(self, model_name, usefast):
             assert_model(response)
 
 
+class TestChatAwq:
+    """Test cases for chat model."""
+
+    @pytest.mark.parametrize(
+        'model_name',
+        ['internlm/internlm2-chat-20b-4bits'],
+    )
+    def test_demo_default(self, model_name):
+        engine_config = TurbomindEngineConfig(model_format='awq')
+        pipe = pipeline('internlm/internlm2-chat-20b-4bits',
+                        backend_config=engine_config)
+        responses = pipe(['Hi, pls intro yourself', 'Shanghai is'])
+        print(responses)
+        for response in responses:
+            assert_model(response.text)
+
+
 class TestBase:
     """Test cases for base model."""
 
     @pytest.mark.parametrize(
         'model_name',
         [
             'internlm/internlm2-7b', 'internlm/internlm2-base-7b',
+            'internlm/internlm2-20b', 'internlm/internlm2-base-20b',
             'internlm/internlm2-1_8b'
         ],
     )
@@ -140,6 +163,7 @@ class TestMMModel:
         'model_name',
         [
             'internlm/internlm-xcomposer2-7b',
+            'internlm/internlm-xcomposer2-7b-4bit'
         ],
     )
     def test_demo_default(self, model_name):
@@ -148,12 +172,16 @@ def test_demo_default(self, model_name):
         # Set `torch_dtype=torch.float16` to load model in float16, otherwise
         # it will be loaded as float32 and might cause OOM Error.
 
-        model = AutoModelForCausalLM.from_pretrained(
-            model_name, torch_dtype=torch.float32,
-            trust_remote_code=True).cuda()
+        if '4bit' in model_name:
+            model = InternLMXComposer2QForCausalLM.from_quantized(
+                model_name, trust_remote_code=True, device='cuda:0').eval()
+        else:
+            model = AutoModelForCausalLM.from_pretrained(
+                model_name, torch_dtype=torch.float32,
+                trust_remote_code=True).cuda()
+
         tokenizer = AutoTokenizer.from_pretrained(model_name,
                                                   trust_remote_code=True)
-
         model = model.eval()
         img_path_list = [
             'tests/panda.jpg',
@@ -175,7 +203,7 @@ def test_demo_default(self, model_name):
                                            do_sample=False)
         print(response)
         assert len(response) != 0
-        assert 'panda' in response
+        assert ' panda' in response
 
         query = '<ImageHere> <ImageHere>请根据图片写一篇作文：我最喜欢的小动物。' \
             + '要求：选准角度，确定立意，明确文体，自拟标题。'
@@ -197,6 +225,7 @@ class TestMMVlModel:
         'model_name',
         [
             'internlm/internlm-xcomposer2-vl-7b',
+            'internlm/internlm-xcomposer2-vl-7b-4bit'
         ],
     )
     def test_demo_default(self, model_name):
@@ -206,8 +235,13 @@ def test_demo_default(self, model_name):
         torch.set_grad_enabled(False)
 
         # init model and tokenizer
-        model = AutoModel.from_pretrained(
-            model_name, trust_remote_code=True).cuda().eval()
+        if '4bit' in model_name:
+            model = InternLMXComposer2QForCausalLM.from_quantized(
+                model_name, trust_remote_code=True, device='cuda:0').eval()
+        else:
+            model = AutoModel.from_pretrained(
+                model_name, trust_remote_code=True).cuda().eval()
+
         tokenizer = AutoTokenizer.from_pretrained(model_name,
                                                   trust_remote_code=True)
 
@@ -223,3 +257,20 @@ def test_demo_default(self, model_name):
         assert len(response) != 0
         assert 'Oscar Wilde' 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'],
+    ]