Merge branch '0.6.2.post1' into bump-version

InternLM · Nov 5, 2024 · 19d44fc · 19d44fc
2 parents 3c32620 + 434195e
commit 19d44fc
Show file tree

Hide file tree

Showing 14 changed files with 894 additions and 40 deletions.
diff --git a/.github/scripts/eval_chat_config.py b/.github/scripts/eval_chat_config.py
@@ -174,7 +174,6 @@
                                 max_out_len=MAX_NEW_TOKENS,
                                 max_seq_len=MAX_SESSION_LEN,
                                 batch_size=128,
-                                concurrency=128,
                                 meta_template=llama2_meta_template,
                                 run_cfg=dict(num_gpus=1),
                                 end_str='[INST]')

diff --git a/.github/workflows/daily_ete_test.yml b/.github/workflows/daily_ete_test.yml
@@ -199,9 +199,9 @@ jobs:
           chmod -R 777 $workdir
 
   test_tools:
-    needs: test_quantization
     if: ${{!cancelled() && (github.event_name == 'schedule' || contains(fromJSON(github.event.inputs.regression_func), 'tools'))}}
     runs-on: [self-hosted, linux-a100]
+    needs: test_quantization
     timeout-minutes: 150
     strategy:
       fail-fast: false

diff --git a/.github/workflows/daily_ete_test_v100.yml b/.github/workflows/daily_ete_test_v100.yml
diff --git a/autotest/config-v100.yaml b/autotest/config-v100.yaml
@@ -0,0 +1,131 @@
+model_path: /nvme/qa_test_models
+dst_path: /nvme/qa_test_models/autotest_model
+log_path: /nvme/qa_test_models/autotest_model/log
+benchmark_path: /nvme/qa_test_models/benchmark-reports
+dataset_path: /nvme/qa_test_models/datasets/ShareGPT_V3_unfiltered_cleaned_split.json
+
+tp_config:
+    internlm-chat-20b: 2
+    internlm2-chat-20b: 2
+    Baichuan2-13B-Chat: 2
+    Mixtral-8x7B-Instruct-v0.1: 2
+    Qwen-VL-Chat: 2
+    llava-v1.5-13b: 2
+    internlm2_5-20b-chat: 2
+    internlm2_5-20b: 2
+    Meta-Llama-3-1-70B-Instruct: 4
+    internlm2_5-7b-chat-1m: 4
+    Qwen2-7B-Instruct-GPTQ-Int4: 2
+    InternVL2-26B: 2
+    InternVL2-40B: 2
+    MiniCPM-V-2_6: 2
+
+turbomind_chat_model:
+    - meta-llama/Meta-Llama-3-1-8B-Instruct
+    - meta-llama/Meta-Llama-3-1-8B-Instruct-AWQ
+    - meta-llama/Meta-Llama-3-8B-Instruct
+    - meta-llama/Meta-Llama-3-8B-Instruct-inner-4bits
+    - internlm/internlm2_5-7b-chat
+    - internlm/internlm2_5-20b-chat
+    - internlm/internlm-xcomposer2d5-7b
+    - OpenGVLab/InternVL2-2B
+    - OpenGVLab/InternVL2-8B
+    - OpenGVLab/InternVL2-26B
+    - OpenGVLab/Mini-InternVL-Chat-2B-V1-5
+    - Qwen/Qwen2-7B-Instruct-AWQ
+    - Qwen/Qwen2-1.5B-Instruct
+    - Qwen/Qwen2.5-7B-Instruct
+    - Qwen/Qwen2-7B-Instruct-GPTQ-Int4
+    - mistralai/Mistral-7B-Instruct-v0.3
+    - THUDM/glm-4-9b-chat
+
+
+pytorch_chat_model:
+    - meta-llama/Meta-Llama-3-8B-Instruct
+    - meta-llama/Meta-Llama-3-1-8B-Instruct
+    - internlm/internlm2_5-7b-chat
+    - internlm/internlm2_5-20b-chat
+    - OpenGVLab/InternVL2-2B
+    - OpenGVLab/InternVL2-4B
+    - OpenGVLab/InternVL2-8B
+    - OpenGVLab/InternVL2-26B
+    - Qwen/Qwen2-1.5B-Instruct
+    - Qwen/Qwen1.5-MoE-A2.7B-Chat
+    - Qwen/Qwen2-VL-2B-Instruct
+    - Qwen/Qwen2-VL-7B-Instruct
+    - google/gemma-2-9b-it
+    - mistralai/Mistral-7B-Instruct-v0.2
+    - THUDM/glm-4v-9b
+    - THUDM/glm-4-9b-chat
+    - microsoft/Phi-3-mini-4k-instruct
+    - deepseek-ai/DeepSeek-V2-Lite-Chat
+
+turbomind_base_model:
+    - internlm/internlm2_5-7b
+    - internlm/internlm2_5-20b
+
+pytorch_base_model:
+    - internlm/internlm2_5-7b
+    - internlm/internlm2_5-20b
+
+vl_model:
+    - OpenGVLab/InternVL2-2B
+    - OpenGVLab/InternVL2-4B
+    - OpenGVLab/InternVL2-8B
+    - OpenGVLab/InternVL2-26B
+    - Qwen/Qwen2-VL-2B-Instruct
+    - Qwen/Qwen2-VL-7B-Instruct
+    - internlm/internlm-xcomposer2d5-7b
+    - THUDM/glm-4v-9b
+    - microsoft/Phi-3-mini-4k-instruct
+
+turbomind_quatization:
+    no_awq:
+        - meta-llama/Meta-Llama-3-1-8B-Instruct
+        - meta-llama/Meta-Llama-3-8B-Instruct
+        - internlm/internlm-xcomposer2d5-7b
+        - OpenGVLab/Mini-InternVL-Chat-2B-V1-5
+        - mistralai/Mistral-7B-Instruct-v0.3
+        - THUDM/glm-4-9b-chat
+    gptq:
+        - internlm/internlm2_5-7b-chat
+    no_kvint4:
+        - openbmb/MiniCPM-V-2_6
+    no_kvint8:
+        - deepseek-ai/DeepSeek-V2-Lite-Chat
+
+pytorch_quatization:
+    awq:
+        - internlm/internlm2_5-7b-chat
+        - internlm/internlm2_5-20b-chat
+        - Qwen/Qwen2-1.5B-Instruct
+    w8a8:
+        - internlm/internlm2_5-7b-chat
+        - internlm/internlm2_5-7b
+    no_kvint4:
+        - OpenGVLab/InternVL2-4B
+        - deepseek-ai/DeepSeek-V2-Lite-Chat
+        - microsoft/Phi-3-mini-4k-instruct
+        - microsoft/Phi-3-vision-128k-instruct
+    no_kvint8:
+        - deepseek-ai/DeepSeek-V2-Lite-Chat
+
+
+longtext_model:
+    - meta-llama/Meta-Llama-3-1-8B-Instruct
+    - meta-llama/Meta-Llama-3-8B-Instruct
+    - meta-llama/Meta-Llama-3-1-70B-Instruct
+    - internlm/internlm2_5-7b-chat-1m
+    - internlm/internlm2-chat-20b
+
+benchmark_model:
+    - meta-llama/Llama-2-7b-chat-hf
+    - meta-llama/Meta-Llama-3-1-8B-Instruct
+    - meta-llama/Meta-Llama-3-8B-Instruct
+    - meta-llama/Meta-Llama-3-1-70B-Instruct
+    - internlm/internlm2_5-7b-chat
+    - internlm/internlm2_5-20b-chat
+    - THUDM/glm-4-9b-chat
+    - mistralai/Mistral-7B-Instruct-v0.3
+    - mistralai/Mixtral-8x7B-Instruct-v0.1
+    - deepseek-ai/DeepSeek-V2-Lite-Chat
diff --git a/autotest/config.yaml b/autotest/config.yaml
@@ -48,6 +48,7 @@ turbomind_chat_model:
     - mistralai/Mistral-7B-Instruct-v0.1
     - mistralai/Mistral-7B-Instruct-v0.2
     - mistralai/Mistral-7B-Instruct-v0.3
+    - mistralai/Mixtral-8x7B-Instruct-v0.1
     - lmdeploy/llama2-chat-7b-w4
     - baichuan-inc/Baichuan2-7B-Chat
     - 01-ai/Yi-6B-Chat
@@ -90,7 +91,6 @@ pytorch_chat_model:
     - mistralai/Mistral-7B-Instruct-v0.1
     - mistralai/Mistral-7B-Instruct-v0.2
     - mistralai/Mixtral-8x7B-Instruct-v0.1
-    - mistralai/Mixtral-8x7B-Instruct-v0.1
     - google/gemma-7b-it
     - google/gemma-2-9b-it
     - deepseek-ai/deepseek-moe-16b-chat