[ci] fix some fail in daily testcase (#3134)

* update * update * update * Update evaluate.yml * Update action_tools.py * update * update * update * update * Update evaluate.yml * update * Update evaluate.yml * Update evaluate.yml * Update evaluate.yml * Update evaluate.yml * update * update * update * update * fix lint
InternLM · Feb 12, 2025 · 188f069 · 188f069
1 parent e91ccf0
commit 188f069
Show file tree

Hide file tree

Showing 12 changed files with 144 additions and 99 deletions.
diff --git a/.github/scripts/action_tools.py b/.github/scripts/action_tools.py
@@ -35,7 +35,7 @@ def run_cmd(cmd_lines: List[str], log_path: str, cwd: str = None):
     cmd_for_run = ' '.join(cmd_lines)
     cmd_for_log = f' {sep}\n'.join(cmd_lines) + '\n'
     with open(log_path, 'w', encoding='utf-8') as file_handler:
-        file_handler.write(f'Command:\n{cmd_for_log}\n')
+        file_handler.write(f'Command: {cmd_for_log}\n')
         file_handler.flush()
         process_res = subprocess.Popen(cmd_for_run, shell=True, cwd=cwd, stdout=file_handler, stderr=file_handler)
         process_res.wait()
@@ -93,17 +93,16 @@ def evaluate(models: List[str], datasets: List[str], workspace: str, evaluate_ty
         print(f'Start evaluating {idx+1}/{num_model} {ori_model} ...')
         model = ori_model.lower()
 
-        opencompass_dir = os.path.abspath(os.environ['OPENCOMPASS_DIR'])
         lmdeploy_dir = os.path.abspath(os.environ['LMDEPLOY_DIR'])
         config_path = os.path.join(lmdeploy_dir, f'.github/scripts/eval_{evaluate_type}_config.py')
-        config_path_new = os.path.join(opencompass_dir, 'configs', 'eval_lmdeploy.py')
+        config_path_new = os.path.join(lmdeploy_dir, 'eval_lmdeploy.py')
         if os.path.exists(config_path_new):
             os.remove(config_path_new)
         shutil.copy(config_path, config_path_new)
 
         cfg = Config.fromfile(config_path_new)
         if not hasattr(cfg, model):
-            logging.error(f'Model {model} not found in configuration file')
+            logging.error(f'Model {model} not in configuration file')
             continue
 
         model_cfg = cfg[model]
@@ -116,13 +115,13 @@ def evaluate(models: List[str], datasets: List[str], workspace: str, evaluate_ty
                 f.write("    if d['reader_cfg'] is not None:\n")
                 f.write("        d['reader_cfg']['test_range'] = '[0:50]'\n")
             if model.startswith('hf'):
-                f.write(f'\nmodels = [ *{model} ]\n')
+                f.write(f'\nmodels = [*{model}]\n')
             else:
-                f.write(f'\nmodels = [ {model} ]\n')
+                f.write(f'\nmodels = [{model}]\n')
 
         work_dir = os.path.join(workspace, model)
         cmd_eval = [
-            f'python3 {opencompass_dir}/run.py {config_path_new} -w {work_dir} --reuse --max-num-workers 8 --dump-eval-details'  # noqa: E501
+            f'opencompass {config_path_new} -w {work_dir} --reuse --max-num-workers 8'  # noqa: E501
         ]
         eval_log = os.path.join(workspace, f'eval.{ori_model}.txt')
         start_time = time.time()
@@ -158,7 +157,7 @@ def evaluate(models: List[str], datasets: List[str], workspace: str, evaluate_ty
         if len(crows_pairs_json) == 1:
             with open(crows_pairs_json[0], 'r') as f:
                 acc = json.load(f)['accuracy']
-                acc = f'{float(acc):.2f}'
+                acc = f'{float(acc):.2f}'  # noqa E231
                 model_results['crows_pairs'] = acc
         logging.info(f'\n{model}\n{model_results}')
         dataset_names = list(model_results.keys())

diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml
@@ -106,7 +106,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r /nvme/qa_test_models/offline_pkg/requirements.txt
       - name: Install lmdeploy
         if: ${{github.event_name == 'schedule' || !inputs.offline_mode}}

diff --git a/.github/workflows/daily_ete_test.yml b/.github/workflows/daily_ete_test.yml
@@ -154,7 +154,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -245,7 +244,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -315,6 +313,8 @@ jobs:
       - name: Test lmdeploy - local testcase
         if: matrix.backend == 'turbomind' && matrix.model == 'llm' && matrix.function == 'local_case'
         run: |
+          pytest autotest/toolchain --alluredir=${{env.REPORT_DIR}} ${{env.COV_PARAM}} || true
+          mv .coverage ${{env.REPORT_DIR}}/.coverage.$(date +'%Y%m%d%H%M%S')
           pytest /local_case/issue_regression --alluredir=${{env.REPORT_DIR}} ${{env.COV_PARAM}}|| true
           mv .coverage ${{env.REPORT_DIR}}/.coverage.$(date +'%Y%m%d%H%M%S')
       - name: Clear workfile
@@ -353,7 +353,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -444,7 +443,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -504,7 +502,6 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -566,7 +563,7 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
+          python3 -m pip install sentence_transformers==2.2.2 --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -577,6 +574,7 @@ jobs:
         run: |
           git clone --depth=1 https://github.com/open-compass/opencompass.git
           cd opencompass
+          cp /nvme/qa_test_models/offline_pkg/requirements-oc.txt requirements/runtime.txt
           python3 -m pip install -e .
           echo "OPENCOMPASS_DIR=$(pwd)" >> $GITHUB_ENV
       - name: Check env

diff --git a/.github/workflows/daily_ete_test_v100.yml b/.github/workflows/daily_ete_test_v100.yml
@@ -136,16 +136,14 @@ jobs:
     timeout-minutes: 180
     env:
       PYTHONPATH: /nvme/qa_test_models/offline_pkg/LLaVA
-      MODELSCOPE_CACHE: /root/modelscope_hub
-      MODELSCOPE_MODULES_CACHE: /root/modelscope_modules
+      MODELSCOPE_CACHE: /nvme/qa_test_models/modelscope_hub
+      MODELSCOPE_MODULES_CACHE: /nvme/qa_test_models/modelscope_modules
     container:
       image: openmmlab/lmdeploy:latest-cu12
       options: "--gpus=all --ipc=host --user root -e PIP_CACHE_DIR=/root/.cache/pip -e NVIDIA_DISABLE_REQUIRE=1 --pull never"
       volumes:
         - /nvme/github-actions/pip-cache:/root/.cache/pip
         - /nvme/github-actions/packages:/root/packages
-        - /nvme/github-actions/modelscope_hub:/root/modelscope_hub
-        - /nvme/github-actions/modelscope_modules:/root/modelscope_modules
         - /nvme/qa_test_models:/nvme/qa_test_models
         - /mnt/shared:/mnt/shared
         - /mnt/187:/mnt/187
@@ -158,7 +156,6 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -220,16 +217,14 @@ jobs:
             function: local_case
     env:
       PYTHONPATH: /nvme/qa_test_models/offline_pkg/LLaVA
-      MODELSCOPE_CACHE: /root/modelscope_hub
-      MODELSCOPE_MODULES_CACHE: /root/modelscope_modules
+      MODELSCOPE_CACHE: /nvme/qa_test_models/modelscope_hub
+      MODELSCOPE_MODULES_CACHE: /nvme/qa_test_models/modelscope_modules
     container:
       image: openmmlab/lmdeploy:latest-cu12
       options: "--gpus=all --ipc=host --user root -e PIP_CACHE_DIR=/root/.cache/pip -e NVIDIA_DISABLE_REQUIRE=1 --pull never"
       volumes:
         - /nvme/github-actions/pip-cache:/root/.cache/pip
         - /nvme/github-actions/packages:/root/packages
-        - /nvme/github-actions/modelscope_hub:/root/modelscope_hub
-        - /nvme/github-actions/modelscope_modules:/root/modelscope_modules
         - /nvme/github-actions/resources/lora:/root/lora
         - /nvme/qa_test_models:/nvme/qa_test_models
         - /mnt/shared:/mnt/shared
@@ -243,7 +238,6 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -341,7 +335,6 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -431,7 +424,6 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -490,7 +482,6 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -550,7 +541,7 @@ jobs:
         run: |
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
+          python3 -m pip install sentence_transformers==2.2.2 --no-deps
           python3 -m pip install -r ${{env.OFFLINE_REQUIREMENTS}}
       - name: Install lmdeploy
         run: |
@@ -560,6 +551,7 @@ jobs:
         run: |
           git clone --depth=1 https://github.com/open-compass/opencompass.git
           cd opencompass
+          cp /nvme/qa_test_models/offline_pkg/requirements-oc.txt requirements/runtime.txt
           python3 -m pip install -e .
           echo "OPENCOMPASS_DIR=$(pwd)" >> $GITHUB_ENV
       - name: Check env

diff --git a/.github/workflows/evaluate.yml b/.github/workflows/evaluate.yml
@@ -17,22 +17,22 @@ on:
         required: true
         description: 'Tested TurboMind models list. eg. [internlm_chat_7b,internlm_chat_7b_w8a16]'
         type: string
-        default: '[turbomind_internlm2_chat_7b, pytorch_internlm2_chat_7b, turbomind_internlm2_5_7b_chat, pytorch_internlm2_5_7b_chat, turbomind_internlm2_5_7b_chat_batch1, turbomind_internlm2_5_7b_chat_batch1_4bits, turbomind_internlm3_8b_instruct, pytorch_internlm3_8b_instruct, turbomind_internlm2_5_20b_chat, pytorch_internlm2_5_20b_chat, turbomind_internlm2_chat_20b, pytorch_internlm2_chat_20b, turbomind_qwen1_5_7b_chat, pytorch_qwen1_5_7b_chat, turbomind_llama3_8b_instruct, pytorch_llama3_8b_instruct, turbomind_llama3_1_8b_instruct, pytorch_llama3_1_8b_instruct, turbomind_qwen2_7b_instruct, pytorch_qwen2_7b_instruct, turbomind_qwen2_5_7b_instruct, pytorch_qwen2_5_7b_instruct, turbomind_llama2_7b_chat, pytorch_qwen1_5_moe_2_7b_chat, pytorch_gemma_2_9b_it, pytorch_gemma_2_27b_it, turbomind_internlm2_chat_7b_4bits, turbomind_internlm2_chat_7b_kvint4, turbomind_internlm2_chat_7b_kvint8, turbomind_internlm2_5_7b_chat_4bits, turbomind_internlm2_5_7b_chat_kvint4, turbomind_internlm2_5_7b_chat_kvint8, pytorch_internlm2_5_7b_chat_w8a8, turbomind_internlm3_8b_instruct_4bits, turbomind_internlm3_8b_instruct_kvint4, turbomind_internlm3_8b_instruct_kvint8, pytorch_internlm3_8b_instruct_w8a8, turbomind_internlm2_5_20b_chat_4bits, turbomind_internlm2_5_20b_chat_kvint4, turbomind_internlm2_5_20b_chat_kvint8, turbomind_llama3_8b_instruct_4bits, turbomind_llama3_8b_instruct_kvint4, turbomind_llama3_1_8b_instruct_4bits, turbomind_llama3_1_8b_instruct_kvint4, turbomind_llama3_1_8b_instruct_kvint8,turbomind_llama3_8b_instruct_kvint8, pytorch_llama3_1_8b_instruct_w8a8, turbomind_qwen2_7b_instruct_4bits, turbomind_qwen2_7b_instruct_kvint4, turbomind_qwen2_7b_instruct_kvint8, pytorch_qwen2_7b_instruct_w8a8, turbomind_qwen2_5_7b_instruct_4bits, turbomind_qwen2_5_7b_instruct_kvint4, turbomind_qwen2_5_7b_instruct_kvint8, pytorch_qwen2_5_7b_instruct_w8a8, turbomind_llama2_7b_chat_4bits, turbomind_llama2_7b_chat_kvint4, turbomind_llama2_7b_chat_kvint8]'
+        default: '[turbomind_internlm2_chat_7b, pytorch_internlm2_chat_7b, turbomind_internlm2_5_7b_chat, pytorch_internlm2_5_7b_chat, turbomind_internlm2_5_7b_chat_batch1, turbomind_internlm2_5_7b_chat_batch1_4bits, turbomind_internlm3_8b_instruct, pytorch_internlm3_8b_instruct, turbomind_internlm2_5_20b_chat, pytorch_internlm2_5_20b_chat, turbomind_internlm2_chat_20b, pytorch_internlm2_chat_20b, turbomind_qwen1_5_7b_chat, pytorch_qwen1_5_7b_chat, turbomind_llama3_8b_instruct, pytorch_llama3_8b_instruct, turbomind_llama3_1_8b_instruct, pytorch_llama3_1_8b_instruct, turbomind_qwen2_7b_instruct, pytorch_qwen2_7b_instruct, turbomind_qwen2_5_7b_instruct, pytorch_qwen2_5_7b_instruct, turbomind_llama2_7b_chat, pytorch_qwen1_5_moe_2_7b_chat, pytorch_gemma_2_9b_it, pytorch_gemma_2_27b_it, turbomind_internlm2_chat_7b_4bits, turbomind_internlm2_chat_7b_kvint4, turbomind_internlm2_chat_7b_kvint8, turbomind_internlm2_5_7b_chat_4bits, turbomind_internlm2_5_7b_chat_kvint4, turbomind_internlm2_5_7b_chat_kvint8, pytorch_internlm2_5_7b_chat_w8a8, turbomind_internlm3_8b_instruct_4bits, turbomind_internlm3_8b_instruct_kvint4, turbomind_internlm3_8b_instruct_kvint8, pytorch_internlm3_8b_instruct_w8a8, turbomind_internlm2_5_20b_chat_4bits, turbomind_internlm2_5_20b_chat_kvint4, turbomind_internlm2_5_20b_chat_kvint8, turbomind_llama3_8b_instruct_4bits, turbomind_llama3_8b_instruct_kvint4, turbomind_llama3_1_8b_instruct_4bits, turbomind_llama3_1_8b_instruct_kvint4, turbomind_llama3_1_8b_instruct_kvint8,turbomind_llama3_8b_instruct_kvint8, pytorch_llama3_1_8b_instruct_w8a8, turbomind_qwen2_7b_instruct_4bits, turbomind_qwen2_7b_instruct_kvint8, pytorch_qwen2_7b_instruct_w8a8, turbomind_qwen2_5_7b_instruct_4bits, turbomind_qwen2_5_7b_instruct_kvint8, pytorch_qwen2_5_7b_instruct_w8a8, turbomind_llama2_7b_chat_4bits, turbomind_llama2_7b_chat_kvint4, turbomind_llama2_7b_chat_kvint8]'
       chat_datasets:
         required: true
         description: 'Tested datasets list. eg. [*bbh_datasets,*ceval_datasets,*cmmlu_datasets,*GaokaoBench_datasets,*gpqa_datasets,*gsm8k_datasets,*hellaswag_datasets,*humaneval_datasets,*ifeval_datasets,*math_datasets,*sanitized_mbpp_datasets,*mmlu_datasets,*nq_datasets,*race_datasets,*TheoremQA_datasets,*triviaqa_datasets,*winogrande_datasets,*crowspairs_datasets]'
         type: string
         default: '[*mmlu_datasets, *gsm8k_datasets, *ifeval_datasets]'
       base_models:
         required: true
-        description: 'Tested TurboMind models list. eg. [turbomind_internlm2_5_7b, turbomind_qwen2_7b]'
+        description: 'Tested TurboMind models list. eg. [turbomind_internlm2_5_7b, turbomind_internlm2_5_7b_4bits, turbomind_internlm2_5_7b_batch1, turbomind_internlm2_5_7b_batch1_4bits, turbomind_qwen2_7b, turbomind_qwen2_5_7b, turbomind_qwen2_5_14b]'
         type: string
-        default: '[turbomind_internlm2_5_7b, turbomind_internlm2_5_7b_4bits, turbomind_internlm2_5_7b_batch1, turbomind_internlm2_5_7b_batch1_4bits, turbomind_qwen2_7b, turbomind_qwen2_5_7b, turbomind_qwen2_5_14b]'
+        default: '[turbomind_internlm2_5_7b, turbomind_internlm2_5_7b_4bits, turbomind_qwen2_7b, turbomind_qwen2_5_7b, turbomind_qwen2_5_14b]'
       baes_datasets:
         required: true
         description: 'Tested datasets list. eg. [*mmlu_datasets, *gsm8k_datasets]'
         type: string
-        default: '[*mmlu_datasets, *gsm8k_datasets, *gpqa_datasets, *winogrande_datasets]'
+        default: '[*race_datasets, *gsm8k_datasets, *gpqa_datasets, *winogrande_datasets]'
       oc_repo_org:
         required: false
         description: 'Tested repository organization name. Default is open-compass/opencompass'
@@ -134,9 +134,7 @@ jobs:
           # manually install flash attn
           # the install packeage from. https://github.com/Dao-AILab/flash-attention/releases
           python3 -m pip install /root/packages/flash_attn-*.whl
-          python3 -m pip install -e /root/packages/AutoAWQ_kernels
-          python3 -m pip install /root/packages/autoawq-*.whl --no-deps
-          python3 -m pip install /root/packages/xformers-*.whl --no-deps
+          python3 -m pip install sentence_transformers==2.2.2 --no-deps
           python3 -m pip install -r /root/models/offline_pkg/requirements.txt
       - name: Install lmdeploy
         if: ${{github.event_name == 'schedule' || !inputs.offline_mode}}
@@ -153,6 +151,7 @@ jobs:
           git clone https://github.com/${{ github.event.inputs.oc_repo_org}}.git
           cd opencompass
           git checkout ${{ github.event.inputs.oc_repo_ref}}
+          cp /root/models/offline_pkg/requirements-oc.txt requirements/runtime.txt
           python3 -m pip install -e .
           echo "OPENCOMPASS_DIR=$(pwd)" >> $GITHUB_ENV
       - name: Check env

diff --git a/autotest/config.yaml b/autotest/config.yaml
@@ -217,11 +217,18 @@ turbomind_quatization:
         - Qwen/Qwen2-7B-Instruct
         - Qwen/Qwen2-7B-Instruct-AWQ
         - Qwen/Qwen2-1.5B-Instruct
+        - Qwen/Qwen1.5-7B-Chat
+        - Qwen/Qwen1.5-4B-Chat-AWQ
+        - Qwen/Qwen1.5-MoE-A2.7B-Chat
+        - Qwen/Qwen-VL-Chat
         - Qwen/Qwen2.5-0.5B-Instruct
         - Qwen/Qwen2.5-7B-Instruct
         - Qwen/Qwen2.5-72B-Instruct
+        - Qwen/Qwen2-VL-2B-Instruct
+        - Qwen/Qwen2-VL-7B-Instruct
         - Qwen/Qwen2-7B-Instruct-GPTQ-Int4
         - allenai/Molmo-7B-D-0924
+        - deepseek-ai/DeepSeek-V2-Lite-Chat
     no_kvint8:
         - deepseek-ai/DeepSeek-V2-Chat
     no_converted:
@@ -273,10 +280,13 @@ pytorch_quatization:
         - Qwen/Qwen2-7B-Instruct
         - Qwen/Qwen2-7B-Instruct-AWQ
         - Qwen/Qwen2-1.5B-Instruct
+        - Qwen/Qwen1.5-7B-Chat
+        - Qwen/Qwen1.5-4B-Chat-AWQ
+        - Qwen/Qwen1.5-MoE-A2.7B-Chat
+        - Qwen/Qwen-VL-Chat
         - Qwen/Qwen2.5-0.5B-Instruct
         - Qwen/Qwen2.5-7B-Instruct
         - Qwen/Qwen2.5-72B-Instruct
-        - Qwen/Qwen2-7B-Instruct-GPTQ-Int4
         - Qwen/Qwen2-VL-2B-Instruct
         - Qwen/Qwen2-VL-7B-Instruct
         - deepseek-ai/DeepSeek-V2-Lite-Chat
@@ -293,6 +303,7 @@ longtext_model:
     - meta-llama/Meta-Llama-3-1-70B-Instruct
     - internlm/internlm2_5-7b-chat-1m
     - internlm/internlm2-chat-20b
+    - internlm/internlm3-8b-instruct
 
 benchmark_model:
     - meta-llama/Llama-2-7b-chat-hf