| { | |
| "model": "/code1/train_logs/merge_lora/thinker/v164-20251211-152343-alignment/iter_0008500", | |
| "model_type": "qwen3_vl", | |
| "model_revision": null, | |
| "task_type": "causal_lm", | |
| "torch_dtype": "bfloat16", | |
| "attn_impl": null, | |
| "new_special_tokens": [], | |
| "num_labels": null, | |
| "problem_type": "single_label_classification", | |
| "rope_scaling": null, | |
| "device_map": null, | |
| "max_memory": {}, | |
| "max_model_len": null, | |
| "local_repo_path": null, | |
| "init_strategy": null, | |
| "template": "qwen3_vl", | |
| "system": null, | |
| "max_length": 50176, | |
| "truncation_strategy": "delete", | |
| "max_pixels": null, | |
| "agent_template": null, | |
| "norm_bbox": null, | |
| "use_chat_template": true, | |
| "padding_free": true, | |
| "padding_side": "right", | |
| "loss_scale": "default", | |
| "sequence_parallel_size": 1, | |
| "response_prefix": null, | |
| "template_backend": "swift", | |
| "dataset": [], | |
| "val_dataset": [], | |
| "split_dataset_ratio": 0.0, | |
| "data_seed": 42, | |
| "dataset_num_proc": 1, | |
| "load_from_cache_file": true, | |
| "dataset_shuffle": true, | |
| "val_dataset_shuffle": false, | |
| "streaming": false, | |
| "interleave_prob": null, | |
| "stopping_strategy": "first_exhausted", | |
| "shuffle_buffer_size": 1000, | |
| "download_mode": "reuse_dataset_if_exists", | |
| "columns": {}, | |
| "strict": false, | |
| "remove_unused_columns": true, | |
| "model_name": null, | |
| "model_author": null, | |
| "custom_dataset_info": [], | |
| "quant_method": null, | |
| "quant_bits": null, | |
| "hqq_axis": null, | |
| "bnb_4bit_compute_dtype": "bfloat16", | |
| "bnb_4bit_quant_type": "nf4", | |
| "bnb_4bit_use_double_quant": true, | |
| "bnb_4bit_quant_storage": null, | |
| "max_new_tokens": null, | |
| "temperature": null, | |
| "top_k": null, | |
| "top_p": null, | |
| "repetition_penalty": null, | |
| "num_beams": 1, | |
| "stream": false, | |
| "stop_words": [], | |
| "logprobs": false, | |
| "top_logprobs": null, | |
| "ckpt_dir": "/code1/train_logs/merge_lora/thinker/v164-20251211-152343-alignment/iter_0008500_mcore", | |
| "lora_modules": [], | |
| "tuner_backend": "peft", | |
| "train_type": "lora", | |
| "adapters": [], | |
| "external_plugins": [], | |
| "seed": 42, | |
| "model_kwargs": {}, | |
| "load_args": false, | |
| "load_data_args": false, | |
| "packing": true, | |
| "packing_length": 50176, | |
| "lazy_tokenize": false, | |
| "cached_dataset": [ | |
| "/data/qwen3vl_train_split_llava665k_part0", | |
| "/data/qwen3vl_train_split_robobrain_part0_160K", | |
| "/data/qwen3vl_train_split_robovqa_part0", | |
| "/data/qwen3vl_train_split_egoplanit_part0_20K", | |
| "/data/qwen3vl_train_split_pixmo-point-570k_part0", | |
| "/data/qwen3vl_train_split_object-ref-340k_part0", | |
| "/data/qwen3vl_train_split_region-ref-340k_part0", | |
| "/data/qwen3vl_train_split_RefSpatial_vacant_qa_231K_part0", | |
| "/data/qwen3vl_train_split_erpa_7k_part0", | |
| "/data/qwen3vl_train_split_imagebench_train_v3_part0" | |
| ], | |
| "custom_register_path": [], | |
| "use_hf": false, | |
| "hub_token": null, | |
| "ddp_timeout": 18000000, | |
| "ddp_backend": null, | |
| "ignore_args_error": false, | |
| "use_swift_lora": false, | |
| "freeze_llm": false, | |
| "freeze_vit": true, | |
| "freeze_aligner": true, | |
| "freeze_parameters": [], | |
| "freeze_parameters_regex": null, | |
| "freeze_parameters_ratio": 0.0, | |
| "trainable_parameters": [], | |
| "trainable_parameters_regex": null, | |
| "adapter_load": null, | |
| "target_modules": [ | |
| "all-linear" | |
| ], | |
| "target_regex": null, | |
| "modules_to_save": [], | |
| "lora_rank": 8, | |
| "lora_alpha": 32, | |
| "lora_dropout": 0.05, | |
| "lora_bias": "none", | |
| "lora_dtype": null, | |
| "use_rslora": false, | |
| "ref_load": null, | |
| "ref_adapter_load": null, | |
| "beta": 0.1, | |
| "rpo_alpha": null, | |
| "reference_free": false, | |
| "label_smoothing": 0.0, | |
| "f_divergence_type": "reverse_kl", | |
| "loss_type": "sigmoid", | |
| "padded_vocab_size": null, | |
| "initialize_embedding": false, | |
| "mlp_padding_free": false, | |
| "dataloader_persistent_workers": true, | |
| "dataloader_prefetch_factor": 10, | |
| "architectures": null, | |
| "llm_architectures": null, | |
| "max_epochs": 2, | |
| "enable_dft_loss": false, | |
| "enable_channel_loss": false, | |
| "original_max_position_embeddings": null, | |
| "partial_rotary_factor": null, | |
| "use_shared_expert_gate": null, | |
| "vit_gradient_checkpointing": true, | |
| "gradient_checkpointing_kwargs": null, | |
| "linear_num_value_heads": null, | |
| "linear_num_key_heads": null, | |
| "linear_key_head_dim": null, | |
| "linear_value_head_dim": null, | |
| "linear_conv_kernel_dim": null, | |
| "layer_types": null, | |
| "mrope_interleaved": null, | |
| "micro_batch_size": 1, | |
| "global_batch_size": 32, | |
| "recompute_granularity": "full", | |
| "recompute_method": "uniform", | |
| "recompute_num_layers": 1, | |
| "recompute_modules": [ | |
| "core_attn" | |
| ], | |
| "use_cpu_initialization": false, | |
| "deterministic_mode": false, | |
| "train_iters": null, | |
| "log_interval": 5, | |
| "tensorboard_dir": null, | |
| "no_masked_softmax_fusion": false, | |
| "no_bias_dropout_fusion": false, | |
| "no_bias_swiglu_fusion": false, | |
| "no_rope_fusion": null, | |
| "no_gradient_accumulation_fusion": false, | |
| "cross_entropy_loss_fusion": true, | |
| "cross_entropy_fusion_impl": "native", | |
| "calculate_per_token_loss": null, | |
| "use_flash_attn": false, | |
| "attention_backend": "flash", | |
| "optimizer": "adam", | |
| "optimizer_cpu_offload": false, | |
| "optimizer_offload_fraction": 1.0, | |
| "use_precision_aware_optimizer": false, | |
| "main_grads_dtype": "fp32", | |
| "main_params_dtype": "fp32", | |
| "exp_avg_dtype": "fp32", | |
| "exp_avg_sq_dtype": "fp32", | |
| "dataloader_type": "cyclic", | |
| "manual_gc": false, | |
| "manual_gc_interval": 0, | |
| "lr": 1e-06, | |
| "lr_decay_style": "cosine", | |
| "lr_decay_iters": null, | |
| "lr_warmup_iters": 0, | |
| "lr_warmup_fraction": 0.05, | |
| "min_lr": 7e-07, | |
| "weight_decay": 0.1, | |
| "clip_grad": 1.0, | |
| "adam_beta1": 0.9, | |
| "adam_beta2": 0.95, | |
| "adam_eps": 1e-08, | |
| "sgd_momentum": 0.9, | |
| "save": "/code1/train_logs/lora/thinker/v300-20251225-024907", | |
| "save_interval": 200, | |
| "no_save_optim": false, | |
| "no_save_rng": false, | |
| "load": "/code1/train_logs/merge_lora/thinker/v164-20251211-152343-alignment/iter_0008500_mcore", | |
| "no_load_optim": false, | |
| "no_load_rng": false, | |
| "finetune": true, | |
| "ckpt_format": "torch_dist", | |
| "no_initialization": true, | |
| "auto_detect_ckpt_format": true, | |
| "exit_on_missing_checkpoint": true, | |
| "distributed_backend": "nccl", | |
| "local_rank": 0, | |
| "use_distributed_optimizer": true, | |
| "tensor_model_parallel_size": 1, | |
| "pipeline_model_parallel_size": 1, | |
| "decoder_first_pipeline_num_layers": null, | |
| "decoder_last_pipeline_num_layers": null, | |
| "sequence_parallel": true, | |
| "context_parallel_size": 1, | |
| "tp_comm_overlap": false, | |
| "overlap_grad_reduce": false, | |
| "overlap_param_gather": false, | |
| "distributed_timeout_minutes": 300000, | |
| "num_layers": null, | |
| "hidden_size": null, | |
| "ffn_hidden_size": null, | |
| "num_attention_heads": null, | |
| "group_query_attention": null, | |
| "num_query_groups": null, | |
| "max_position_embeddings": null, | |
| "position_embedding_type": null, | |
| "mrope_section": null, | |
| "rotary_base": null, | |
| "rotary_percent": 1.0, | |
| "rotary_interleaved": null, | |
| "normalization": "RMSNorm", | |
| "norm_epsilon": null, | |
| "swiglu": null, | |
| "untie_embeddings_and_output_weights": null, | |
| "disable_bias_linear": null, | |
| "add_qkv_bias": null, | |
| "attention_dropout": null, | |
| "hidden_dropout": 0.0, | |
| "kv_channels": null, | |
| "qk_layernorm": null, | |
| "transformer_impl": "transformer_engine", | |
| "num_experts": null, | |
| "moe_layer_freq": null, | |
| "moe_ffn_hidden_size": null, | |
| "moe_shared_expert_intermediate_size": null, | |
| "moe_router_topk": null, | |
| "moe_router_pre_softmax": null, | |
| "moe_router_dtype": "fp32", | |
| "moe_router_score_function": null, | |
| "moe_router_bias_update_rate": 0.001, | |
| "moe_router_enable_expert_bias": null, | |
| "moe_router_topk_scaling_factor": null, | |
| "moe_router_load_balancing_type": null, | |
| "expert_model_parallel_size": 1, | |
| "expert_tensor_parallel_size": 1, | |
| "moe_token_dispatcher_type": null, | |
| "moe_enable_deepep": false, | |
| "moe_grouped_gemm": false, | |
| "moe_permute_fusion": false, | |
| "moe_aux_loss_coeff": 0.0, | |
| "moe_z_loss_coeff": null, | |
| "moe_expert_capacity_factor": null, | |
| "moe_shared_expert_overlap": false, | |
| "moe_layer_recompute": false, | |
| "moe_token_drop_policy": null, | |
| "multi_latent_attention": null, | |
| "q_lora_rank": null, | |
| "kv_lora_rank": null, | |
| "qk_head_dim": null, | |
| "qk_pos_emb_head_dim": null, | |
| "fp8_format": null, | |
| "fp8_recipe": "delayed", | |
| "fp8_amax_history_len": 1024, | |
| "fp8_amax_compute_algo": "max", | |
| "fp8_param_gather": false, | |
| "fp16": null, | |
| "bf16": null, | |
| "apply_query_key_layer_scaling": null, | |
| "attention_softmax_in_fp32": true, | |
| "log_params_norm": false, | |
| "log_throughput": false, | |
| "tensorboard_log_interval": 1, | |
| "tensorboard_queue_size": 50, | |
| "log_timers_to_tensorboard": true, | |
| "no_log_learning_rate_to_tensorboard": false, | |
| "log_validation_ppl_to_tensorboard": true, | |
| "log_memory_to_tensorboard": true, | |
| "logging_level": null, | |
| "wandb_project": null, | |
| "wandb_exp_name": null, | |
| "wandb_save_dir": null, | |
| "eval_iters": -1, | |
| "eval_interval": null, | |
| "seq_length": 50176, | |
| "num_workers": 8, | |
| "extra_megatron_kwargs": null, | |
| "add_version": true, | |
| "rank": 0, | |
| "global_world_size": 16, | |
| "local_world_size": 8, | |
| "model_suffix": "iter_0008500", | |
| "model_info": "ModelInfo(model_type='qwen3_vl', model_dir='/code1/train_logs/merge_lora/thinker/v164-20251211-152343-alignment/iter_0008500', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling={'mrope_interleaved': True, 'mrope_section': [24, 20, 20], 'rope_type': 'default'}, is_moe_model=False, config=None, task_type='causal_lm', num_labels=None)", | |
| "model_meta": "ModelMeta(model_type='qwen3_vl', model_groups=[], template='qwen3_vl', get_function=<function get_model_tokenizer_qwen3_vl at 0x70d9ef09f880>, model_arch=MultiModelKeys(arch_name='qwen3_vl', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model'], aligner=['model.visual.merger', 'model.visual.deepstack_merger_list'], vision_tower=['model.visual'], generator=[]), architectures=['Qwen3VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.57.0.dev', 'qwen_vl_utils>=0.0.14', 'decord'], tags=['vision', 'video'])", | |
| "model_dir": "/code1/train_logs/merge_lora/thinker/v164-20251211-152343-alignment/iter_0008500", | |
| "hub": "<class 'swift.hub.hub.MSHub'>", | |
| "megatron_model_meta": "MMGPTMegatronModelMeta(megatron_model_type='qwen3_vl', model_types=['qwen3_vl', 'qwen3_moe_vl'], convert_mcore2hf=<function convert_mcore2hf_qwen3_vl at 0x70d988611bc0>, convert_hf2mcore=<function convert_hf2mcore_qwen3_vl at 0x70d988611440>, model_cls=<class 'swift.megatron.model.mm_gpt.qwen3_vl.Qwen3VLGPTModel'>, convert_hf_config=<function convert_gpt_hf_config at 0x70d9885da0c0>, get_transformer_layer_spec=None, model_provider=<function model_provider at 0x70d9885d91c0>, visual_cls=<class 'swift.megatron.model.mm_gpt.qwen3_vl.Qwen3VL_Vit'>, extra_args_provider=None)" | |
| } | |