v1.1

args.json

@@ -25,8 +25,7 @@
   "use_chat_template": true,
   "template_backend": "swift",
   "dataset": [
-    "/mnt/petrelfs/liuhongbo/movie_bench/shotdeck_data/sft/img_train_v5_without_reasoning.json",
-    "/mnt/petrelfs/liuhongbo/movie_bench/process_video/vid_bench/train_v3_rl_s2_remove_duplicate_refine_2sft.json"
+    "/mnt/petrelfs/liuhongbo/shotdeck/ShotQA/sft_merged.json"
   ],
   "val_dataset": [],
   "split_dataset_ratio": 0.0,
@@ -85,22 +84,22 @@
   "custom_register_path": [],
   "ignore_args_error": false,
   "use_swift_lora": false,
-  "output_dir": "/mnt/petrelfs/liuhongbo/ms-swift/output/main/sft_7B_without_reasoning_wvideo/v1-20250623-155054",
+  "output_dir": "/mnt/petrelfs/liuhongbo/ms-swift/output/sft_7B_v1.1/v3-20250901-131037",
   "overwrite_output_dir": false,
   "do_train": false,
   "do_eval": false,
   "do_predict": false,
   "eval_strategy": "steps",
   "prediction_loss_only": false,
-  "per_device_train_batch_size": 1,
-  "per_device_eval_batch_size": 1,
+  "per_device_train_batch_size": 16,
+  "per_device_eval_batch_size": 16,
   "per_gpu_train_batch_size": null,
   "per_gpu_eval_batch_size": null,
-  "gradient_accumulation_steps": 16,
+  "gradient_accumulation_steps": 2,
   "eval_accumulation_steps": null,
   "eval_delay": 0,
   "torch_empty_cache_steps": null,
-  "learning_rate": 5e-06,
+  "learning_rate": 1e-06,
   "weight_decay": 0.1,
   "adam_beta1": 0.9,
   "adam_beta2": 0.95,
@@ -115,13 +114,13 @@
   "log_level": "passive",
   "log_level_replica": "warning",
   "log_on_each_node": true,
-  "logging_dir": "/mnt/petrelfs/liuhongbo/ms-swift/output/main/sft_7B_without_reasoning_wvideo/v1-20250623-155054/runs",
+  "logging_dir": "/mnt/petrelfs/liuhongbo/ms-swift/output/sft_7B_v1.1/v3-20250901-131037/runs",
   "logging_strategy": "steps",
   "logging_first_step": true,
   "logging_steps": 5,
   "logging_nan_inf_filter": true,
   "save_strategy": "steps",
-  "save_steps": 100.0,
+  "save_steps": 1000.0,
   "save_total_limit": 1,
   "save_safetensors": true,
   "save_on_each_node": false,
@@ -145,7 +144,7 @@
   "tpu_metrics_debug": false,
   "debug": null,
   "dataloader_drop_last": false,
-  "eval_steps": 100.0,
+  "eval_steps": 1000.0,
   "dataloader_num_workers": 4,
   "dataloader_prefetch_factor": null,
   "past_index": -1,
@@ -233,7 +232,6 @@
   "include_for_metrics": [],
   "eval_do_concat_batches": true,
   "fp16_backend": "auto",
-  "evaluation_strategy": "steps",
   "push_to_hub_model_id": null,
   "push_to_hub_organization": null,
   "push_to_hub_token": null,
@@ -246,8 +244,6 @@
   "torch_compile": false,
   "torch_compile_backend": null,
   "torch_compile_mode": null,
-  "dispatch_batches": null,
-  "split_batches": null,
   "include_tokens_per_second": false,
   "include_num_input_tokens_seen": false,
   "neftune_noise_alpha": null,
@@ -361,12 +357,13 @@
   "metric": null,
   "zero_hpz_partition_size": null,
   "rank": 0,
-  "global_world_size": 4,
-  "local_world_size": 4,
+  "global_world_size": 8,
+  "local_world_size": 8,
   "model_suffix": "Qwen2.5-VL-7B-Instruct",
   "model_info": "ModelInfo(model_type='qwen2_5_vl', model_dir='/mnt/petrelfs/liuhongbo/Qwen2.5-VL-7B-Instruct', torch_dtype=torch.bfloat16, max_model_len=128000, quant_method=None, quant_bits=None, rope_scaling={'type': 'default', 'mrope_section': [16, 24, 24], 'rope_type': 'default'}, config=None, task_type='causal_lm', num_labels=None)",
-  "model_meta": "ModelMeta(model_type='qwen2_5_vl', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen2_5_vl', get_function=<function get_model_tokenizer_qwen2_5_vl at 0x7f5197320670>, model_arch='qwen2_vl', architectures=['Qwen2_5_VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.49', 'qwen_vl_utils>=0.0.6', 'decord'], tags=[])",
+  "model_meta": "ModelMeta(model_type='qwen2_5_vl', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen2_5_vl', get_function=<function get_model_tokenizer_qwen2_5_vl at 0x7f35ea8f7d00>, model_arch='qwen2_vl', architectures=['Qwen2_5_VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.49', 'qwen_vl_utils>=0.0.6', 'decord'], tags=[])",
   "model_dir": "/mnt/petrelfs/liuhongbo/Qwen2.5-VL-7B-Instruct",
   "hub": "<class 'swift.hub.hub.MSHub'>",
-  "training_args": "Seq2SeqTrainingArguments(output_dir='/mnt/petrelfs/liuhongbo/ms-swift/output/main/sft_7B_without_reasoning_wvideo/v1-20250623-155054', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.NO: 'no'>, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=16, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=5e-06, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=1.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/mnt/petrelfs/liuhongbo/ms-swift/output/main/sft_7B_without_reasoning_wvideo/v1-20250623-155054/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=5, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=100, save_total_limit=1, save_safetensors=True, save_on_each_node=False, save_only_model=True, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=100, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/mnt/petrelfs/liuhongbo/ms-swift/output/main/sft_7B_without_reasoning_wvideo/v1-20250623-155054', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, tp_size=0, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['wandb'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=None, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy=<IntervalStrategy.NO: 'no'>, push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, metric_warmup_step=0, fsdp_num=1, acc_steps=1, eval_use_evalscope=False, eval_datasets=[], eval_limit=None, eval_datasets_args=None, eval_generation_config=None, train_type='full', optimizer=None, local_repo_path=None, galore_config=None)"
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/mnt/petrelfs/liuhongbo/ms-swift/output/sft_7B_v1.1/v3-20250901-131037', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.NO: 'no'>, prediction_loss_only=False, per_device_train_batch_size=16, per_device_eval_batch_size=16, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=2, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=1e-06, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=1.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/mnt/petrelfs/liuhongbo/ms-swift/output/sft_7B_v1.1/v3-20250901-131037/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=5, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=1000, save_total_limit=1, save_safetensors=True, save_on_each_node=False, save_only_model=True, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=1000, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/mnt/petrelfs/liuhongbo/ms-swift/output/sft_7B_v1.1/v3-20250901-131037', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, tp_size=0, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['wandb'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=None, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, metric_warmup_step=0, fsdp_num=1, acc_steps=1, eval_use_evalscope=False, eval_datasets=[], eval_limit=None, eval_datasets_args=None, eval_generation_config=None, train_type='full', optimizer=None, local_repo_path=None, galore_config=None)"
 }

config.json

@@ -31,7 +31,7 @@
   "sliding_window": 32768,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.50.3",
+  "transformers_version": "4.51.3",
   "use_cache": false,
   "use_sliding_window": false,
   "video_token_id": 151656,

generation_config.json

@@ -10,5 +10,5 @@
   "temperature": 0.1,
   "top_k": 1,
   "top_p": 0.001,
-  "transformers_version": "4.50.3"
+  "transformers_version": "4.51.3"
 }

model-00001-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e092af04e4b1cf0427712eb6c65a748026ac404d8c3a1fdc64f9d362fce7ecb5
+oid sha256:0740ea992528b84d89247ad9a3544d688d74596cd6a7cf6cd1c587016b93c458
 size 4968243304

model-00002-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c427fc3ed4ebf47d1ad246b11321e255f7f4685cd15ddc3efae4f0d5d90c7ac6
+oid sha256:d38dc6a2731aeb341bb8ae055692881a48ef94e0b2ab6d29d8d4608e9e2f620d
 size 4991495816

model-00003-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3211bb0a7b56a4a330b102d210e31fe084a5f8d721a030e8940c7dfaf4a1e2c4
+oid sha256:4dd6038d4e15af8b097d2cc7259fa5af2e8310546d268df3dd64f285e9612fef
 size 4932751040

model-00004-of-00004.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dd744074338b4c2b7b9d6ad3628d03e21bc507498b86c842a1cfb0e02c0836af
+oid sha256:e9bcb75184abd54cdd8f38315cb7ad67daee695167ce0c68b5a72ed3e70f64af
 size 1691924384

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f7dec8e5ecd03f7b60cd0b0dfcaa6013458cf9e6817c81c5c391d78b95ee914
-size 8312
+oid sha256:5ec30db82d58e502de1d84aeaa00ff3be6dcdbab98c36a7d211221c3f43dc348
+size 9976