[Sync] update model configs (#574)

.gitignore

@@ -89,3 +89,4 @@ docs/zh_cn/_build/
 
 # sft config ignore list
 configs/sft_cfg/*B_*
+configs/cky/

configs/datasets/GLUE_MRPC/GLUE_MRPC_ppl_96564c.py

@@ -22,8 +22,8 @@ MRPC_infer_cfg = dict(
         },
         ice_token='</E>',
     ),
-    retriever=dict(type=FixKRetriever),
-    inferencer=dict(type=PPLInferencer, fix_id_list=[0, 1, 2, 3, 4]))
+    retriever=dict(type=FixKRetriever, fix_id_list=[0, 1, 2, 3, 4]),
+    inferencer=dict(type=PPLInferencer))
 
 MRPC_eval_cfg = dict(evaluator=dict(type=AccEvaluator), )
 

configs/datasets/humaneval/humaneval_gen_4a6eef.py

+from opencompass.openicl.icl_prompt_template import PromptTemplate
+from opencompass.openicl.icl_retriever import ZeroRetriever
+from opencompass.openicl.icl_inferencer import GenInferencer
+from opencompass.datasets import HFDataset, HumanEvaluator, humaneval_postprocess
+
+humaneval_reader_cfg = dict(
+    input_columns=['prompt'], output_column='task_id', train_split='test')
+
+# TODO: allow empty output-column
+humaneval_infer_cfg = dict(
+    prompt_template=dict(
+        type=PromptTemplate,
+        template=dict(round=[
+            dict(
+                role='HUMAN',
+                prompt='Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\nCreate a Python script for this problem:\n{prompt}\n\n### Response:\n'),
+        ])),
+    retriever=dict(type=ZeroRetriever),
+    inferencer=dict(type=GenInferencer, max_out_len=512))
+
+humaneval_eval_cfg = dict(
+    evaluator=dict(type=HumanEvaluator),
+    pred_role='BOT',
+    k=[1, 10, 100],  # the parameter only for humaneval
+    pred_postprocessor=dict(type=humaneval_postprocess),
+)
+
+humaneval_datasets = [
+    dict(
+        type=HFDataset,
+        path='openai_humaneval',
+        reader_cfg=humaneval_reader_cfg,
+        infer_cfg=humaneval_infer_cfg,
+        eval_cfg=humaneval_eval_cfg)
+]

configs/models/aquila/hf_aquila2_34b.py

+from opencompass.models import HuggingFaceCausalLM
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquila2-34b-hf',
+        path="BAAI/Aquila2-34B",
+        tokenizer_path='BAAI/Aquila2-34B',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=2, num_procs=1),
+    )
+]

configs/models/aquila/hf_aquila2_7b.py

+from opencompass.models import HuggingFaceCausalLM
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquila2-7b-hf',
+        path="BAAI/Aquila2-7B",
+        tokenizer_path='BAAI/Aquila2-7B',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]

configs/models/aquila/hf_aquilachat2_34b.py

+from opencompass.models import HuggingFaceCausalLM
+
+_meta_template = dict(
+    round=[
+        dict(role='HUMAN', begin='### Human: ', end='\n'),
+        dict(role='BOT', begin='### Assistant: ', end='</s>', generate=True),
+    ],
+    eos_token_id=100007,
+)
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquilachat2-34b-hf',
+        path="BAAI/AquilaChat2-34B",
+        tokenizer_path='BAAI/AquilaChat2-34B',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        meta_template=_meta_template,
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=2, num_procs=1),
+    )
+]

configs/models/aquila/hf_aquilachat2_34b_16k.py

+from opencompass.models import HuggingFaceCausalLM
+
+_meta_template = dict(
+    begin='###',
+    round=[
+        dict(role='HUMAN', begin='Human: ', end='###'),
+        dict(role='BOT', begin='Assistant: ', end='</s>', generate=True),
+    ],
+    eos_token_id=100007,
+)
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquilachat2-34b-16k-hf',
+        path="BAAI/AquilaChat2-34B-16K",
+        tokenizer_path='BAAI/AquilaChat2-34B-16K',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        meta_template=_meta_template,
+        max_out_len=100,
+        max_seq_len=4096,
+        batch_size=8,
+        run_cfg=dict(num_gpus=2, num_procs=1),
+    )
+]

configs/models/aquila/hf_aquilachat2_7b.py

+from opencompass.models import HuggingFaceCausalLM
+
+_meta_template = dict(
+    round=[
+        dict(role='HUMAN', begin='<|startofpiece|>', end=''),
+        dict(role='BOT', begin='<|endofpiece|>', end='</s>', generate=True),
+    ],
+    eos_token_id=2,
+)
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquilachat2-7b-hf',
+        path="BAAI/AquilaChat2-7B",
+        tokenizer_path='BAAI/AquilaChat2-7B',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        meta_template=_meta_template,
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]

configs/models/aquila/hf_aquilachat2_7b_16k.py

+from opencompass.models import HuggingFaceCausalLM
+
+_meta_template = dict(
+    begin='###',
+    round=[
+        dict(role='HUMAN', begin='Human: ', end='###'),
+        dict(role='BOT', begin='Assistant: ', end='</s>', generate=True),
+    ],
+    eos_token_id=100007,
+)
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='aquilachat2-7b-16k-hf',
+        path="BAAI/AquilaChat2-7B-16K",
+        tokenizer_path='BAAI/AquilaChat2-7B-16K',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+            use_fast=False,
+        ),
+        meta_template=_meta_template,
+        max_out_len=100,
+        max_seq_len=4096,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]

configs/models/chatglm/hf_chatglm2_6b.py

@@ -7,15 +7,18 @@ models = [
         abbr='chatglm2-6b-hf',
         path='THUDM/chatglm2-6b',
         tokenizer_path='THUDM/chatglm2-6b',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
         tokenizer_kwargs=dict(
            padding_side='left',
            truncation_side='left',
            trust_remote_code=True,
         ),
         max_out_len=100,
-        max_seq_len=2048,
+        max_seq_len=4096,
         batch_size=8,
-        model_kwargs=dict(trust_remote_code=True, device_map='auto', revision='a6d54fac46dff2db65d53416c207a4485ca6bd40'),
         run_cfg=dict(num_gpus=1, num_procs=1),
     )
 ]

configs/models/chatglm/hf_chatglm3_6b.py

+from opencompass.models import HuggingFaceChatGLM3
+
+api_meta_template = dict(
+    round=[
+        dict(role='HUMAN', api_role='HUMAN'),
+        dict(role='BOT', api_role='BOT', generate=True),
+    ]
+)
+
+models = [
+    dict(
+        type=HuggingFaceChatGLM3,
+        abbr='chatglm3-6b-hf',
+        path='THUDM/chatglm3-6b',
+        tokenizer_path='THUDM/chatglm3-6b',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+        ),
+        meta_template=api_meta_template,
+        max_out_len=100,
+        max_seq_len=4096,
+        batch_size=1,
+        run_cfg=dict(num_gpus=1, num_procs=1)
+    )
+]

configs/models/chatglm/hf_chatglm3_6b_base.py

+from opencompass.models import HuggingFace
+
+
+models = [
+    dict(
+        type=HuggingFace,
+        abbr='chatglm3-6b-base-hf',
+        path='THUDM/chatglm3-6b-base',
+        tokenizer_path='THUDM/chatglm3-6b-base',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
+        tokenizer_kwargs=dict(
+           padding_side='left',
+           truncation_side='left',
+           trust_remote_code=True,
+        ),
+        max_out_len=100,
+        max_seq_len=4096,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]

configs/models/chatglm/hf_chatglm_6b.py

@@ -7,6 +7,10 @@ models = [
         abbr='chatglm-6b-hf',
         path='THUDM/chatglm-6b',
         tokenizer_path='THUDM/chatglm-6b',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
         tokenizer_kwargs=dict(
             padding_side='left',
             truncation_side='left',
@@ -15,7 +19,6 @@ models = [
         max_out_len=100,
         max_seq_len=2048,
         batch_size=8,
-        model_kwargs=dict(trust_remote_code=True, device_map='auto', revision='1d240ba371910e9282298d4592532d7f0f3e9f3e'),
         run_cfg=dict(num_gpus=1, num_procs=1),
     )
 ]

configs/models/hf_internlm/hf_internlm_20b.py

+from opencompass.models import HuggingFaceCausalLM
+
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='internlm-20b-hf',
+        path="internlm/internlm-20b",
+        tokenizer_path='internlm/internlm-20b',
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            use_fast=False,
+            trust_remote_code=True,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        model_kwargs=dict(trust_remote_code=True, device_map='auto'),
+        run_cfg=dict(num_gpus=2, num_procs=1),
+    )
+]

configs/models/hf_internlm/hf_internlm_7b.py

@@ -7,6 +7,10 @@ models = [
         abbr='internlm-7b-hf',
         path="internlm/internlm-7b",
         tokenizer_path='internlm/internlm-7b',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
         tokenizer_kwargs=dict(
             padding_side='left',
             truncation_side='left',
@@ -16,7 +20,6 @@ models = [
         max_out_len=100,
         max_seq_len=2048,
         batch_size=8,
-        model_kwargs=dict(trust_remote_code=True, device_map='auto'),
         run_cfg=dict(num_gpus=1, num_procs=1),
     )
 ]

configs/models/hf_internlm/hf_internlm_chat_20b.py

+from opencompass.models import HuggingFaceCausalLM
+
+
+_meta_template = dict(
+    round=[
+        dict(role='HUMAN', begin='<|User|>:', end='<eoh>\n'),
+        dict(role='BOT', begin='<|Bot|>:', end='<eoa>\n', generate=True),
+    ],
+)
+
+models = [
+    dict(
+        type=HuggingFaceCausalLM,
+        abbr='internlm-chat-20b-hf',
+        path="internlm/internlm-chat-20b",
+        tokenizer_path='internlm/internlm-chat-20b',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            use_fast=False,
+            trust_remote_code=True,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        meta_template=_meta_template,
+        run_cfg=dict(num_gpus=2, num_procs=1),
+    )
+]

configs/models/hf_internlm/hf_internlm_chat_7b.py

@@ -14,21 +14,20 @@ models = [
         abbr='internlm-chat-7b-hf',
         path="internlm/internlm-chat-7b",
         tokenizer_path='internlm/internlm-chat-7b',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
         tokenizer_kwargs=dict(
             padding_side='left',
             truncation_side='left',
             use_fast=False,
             trust_remote_code=True,
-            revision="1a6328795c6e207904e1eb58177e03ad24ae06f3"
         ),
         max_out_len=100,
         max_seq_len=2048,
         batch_size=8,
         meta_template=_meta_template,
-        model_kwargs=dict(
-            trust_remote_code=True,
-            device_map='auto',
-            revision="1a6328795c6e207904e1eb58177e03ad24ae06f3"),
         run_cfg=dict(num_gpus=1, num_procs=1),
     )
 ]

configs/models/hf_internlm/hf_internlm_chat_7b_8k.py

@@ -14,6 +14,10 @@ models = [
         abbr='internlm-chat-7b-8k-hf',
         path="internlm/internlm-chat-7b-8k",
         tokenizer_path='internlm/internlm-chat-7b-8k',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+        ),
         tokenizer_kwargs=dict(
             padding_side='left',
             truncation_side='left',
@@ -24,7 +28,6 @@ models = [
         max_seq_len=2048,
         batch_size=8,
         meta_template=_meta_template,
-        model_kwargs=dict(trust_remote_code=True, device_map='auto'),
         run_cfg=dict(num_gpus=1, num_procs=1),
     )
 ]

configs/models/lingowhale/hf_lingowhale_8b.py

+from opencompass.models import HuggingFace
+
+
+models = [
+    dict(
+        type=HuggingFace,
+        abbr='lingowhale-8b-hf',
+        path='deeplang-ai/LingoWhale-8B',
+        tokenizer_path='deeplang-ai/LingoWhale-8B',
+        model_kwargs=dict(
+            trust_remote_code=True,
+            device_map='auto',
+            torch_dtype='auto',
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]

configs/models/mistral/hf_mistral_7b.py

+from opencompass.models import HuggingFaceCausalLM
+
+
+models = [
+    dict(
+        abbr='mistral-7b-v0.1-hf',
+        type=HuggingFaceCausalLM,
+        path='mistralai/Mistral-7B-v0.1',
+        tokenizer_path='mistralai/Mistral-7B-v0.1',
+        model_kwargs=dict(
+            device_map='auto',
+            trust_remote_code=True,
+        ),
+        tokenizer_kwargs=dict(
+            padding_side='left',
+            truncation_side='left',
+            trust_remote_code=True,
+        ),
+        max_out_len=100,
+        max_seq_len=2048,
+        batch_size=8,
+        run_cfg=dict(num_gpus=1, num_procs=1),
+    )
+]