Upload args.json with huggingface_hub

args.json

@@ -0,0 +1,461 @@
+{
+  "model": "Qwen3-4B-Thinking-2507",
+  "model_type": "qwen3_thinking",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": null,
+  "new_special_tokens": [],
+  "num_labels": null,
+  "problem_type": null,
+  "rope_scaling": null,
+  "device_map": null,
+  "max_memory": {},
+  "max_model_len": null,
+  "local_repo_path": null,
+  "init_strategy": null,
+  "template": "qwen3_thinking",
+  "system": null,
+  "max_length": 8192,
+  "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": 8,
+  "load_from_cache_file": false,
+  "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": "/workspace/Qwen3-4B-Thinking-2507-mcore",
+  "lora_modules": [],
+  "tuner_backend": "peft",
+  "train_type": "full",
+  "adapters": [],
+  "external_plugins": [],
+  "seed": 42,
+  "model_kwargs": {},
+  "load_args": false,
+  "load_data_args": false,
+  "packing": true,
+  "packing_length": 8192,
+  "lazy_tokenize": false,
+  "cached_dataset": [
+    "my_data_output_only_cached"
+  ],
+  "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,
+  "rlhf_type": null,
+  "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": null,
+  "desirable_weight": 1.0,
+  "undesirable_weight": 1.0,
+  "calculate_KL": null,
+  "center_rewards_coefficient": null,
+  "padded_vocab_size": 151936,
+  "initialize_embedding": false,
+  "mlp_padding_free": false,
+  "dataloader_persistent_workers": true,
+  "dataloader_prefetch_factor": 10,
+  "architectures": "Qwen3ForCausalLM",
+  "llm_architectures": null,
+  "max_epochs": 4,
+  "enable_dft_loss": false,
+  "enable_channel_loss": false,
+  "original_max_position_embeddings": null,
+  "partial_rotary_factor": null,
+  "use_shared_expert_gate": false,
+  "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": false,
+  "micro_batch_size": 4,
+  "global_batch_size": 128,
+  "recompute_granularity": "selective",
+  "recompute_method": null,
+  "recompute_num_layers": null,
+  "recompute_modules": [
+    "core_attn"
+  ],
+  "use_cpu_initialization": false,
+  "deterministic_mode": false,
+  "train_iters": null,
+  "log_interval": 1,
+  "tensorboard_dir": "/workspace/Qwen3-4B-Thinking-2507-mcore-affine-4b/v0-20251113-132445/runs",
+  "no_masked_softmax_fusion": false,
+  "no_bias_dropout_fusion": false,
+  "no_bias_swiglu_fusion": false,
+  "no_rope_fusion": false,
+  "no_gradient_accumulation_fusion": false,
+  "cross_entropy_loss_fusion": true,
+  "cross_entropy_fusion_impl": "native",
+  "calculate_per_token_loss": true,
+  "use_flash_attn": false,
+  "attention_backend": "flash",
+  "optimizer": "adam",
+  "optimizer_cpu_offload": false,
+  "optimizer_offload_fraction": 1.0,
+  "use_precision_aware_optimizer": true,
+  "main_grads_dtype": "bf16",
+  "main_params_dtype": "fp32",
+  "exp_avg_dtype": "bf16",
+  "exp_avg_sq_dtype": "bf16",
+  "dataloader_type": "cyclic",
+  "manual_gc": false,
+  "manual_gc_interval": 0,
+  "lr": 3e-05,
+  "lr_decay_style": "cosine",
+  "lr_decay_iters": null,
+  "lr_warmup_iters": 0,
+  "lr_warmup_fraction": 0.05,
+  "min_lr": 3e-06,
+  "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": "/workspace/Qwen3-4B-Thinking-2507-mcore-affine-4b/v0-20251113-132445",
+  "save_interval": 1490,
+  "save_retain_interval": null,
+  "no_save_optim": true,
+  "no_save_rng": true,
+  "load": "/workspace/Qwen3-4B-Thinking-2507-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,
+  "async_save": false,
+  "use_persistent_ckpt_worker": false,
+  "ckpt_fully_parallel_load": false,
+  "ckpt_assume_constant_structure": false,
+  "distributed_backend": "nccl",
+  "local_rank": 0,
+  "use_distributed_optimizer": true,
+  "tensor_model_parallel_size": 4,
+  "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_per_virtual_pipeline_stage": null,
+  "num_virtual_stages_per_pipeline_rank": null,
+  "microbatch_group_size_per_virtual_pipeline_stage": null,
+  "pipeline_model_parallel_layout": null,
+  "num_layers": 36,
+  "hidden_size": 2560,
+  "ffn_hidden_size": 9728,
+  "num_attention_heads": 32,
+  "group_query_attention": true,
+  "num_query_groups": 8,
+  "max_position_embeddings": 262144,
+  "position_embedding_type": "rope",
+  "mrope_section": null,
+  "rotary_base": 5000000,
+  "rotary_percent": 1.0,
+  "rotary_interleaved": false,
+  "normalization": "RMSNorm",
+  "norm_epsilon": 1e-06,
+  "swiglu": true,
+  "untie_embeddings_and_output_weights": false,
+  "disable_bias_linear": true,
+  "add_qkv_bias": false,
+  "attention_dropout": 0.0,
+  "hidden_dropout": 0.0,
+  "kv_channels": 128,
+  "qk_layernorm": true,
+  "transformer_impl": "transformer_engine",
+  "num_experts": null,
+  "moe_layer_freq": "1",
+  "moe_ffn_hidden_size": null,
+  "moe_shared_expert_intermediate_size": null,
+  "moe_router_topk": 2,
+  "moe_router_pre_softmax": false,
+  "moe_router_dtype": "fp32",
+  "moe_router_score_function": "softmax",
+  "moe_router_bias_update_rate": 0.001,
+  "moe_router_enable_expert_bias": false,
+  "moe_router_topk_scaling_factor": null,
+  "moe_router_load_balancing_type": "aux_loss",
+  "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_shared_expert_overlap": false,
+  "moe_layer_recompute": false,
+  "moe_expert_capacity_factor": null,
+  "moe_pad_expert_input_to_capacity": false,
+  "moe_token_drop_policy": null,
+  "multi_latent_attention": false,
+  "q_lora_rank": null,
+  "kv_lora_rank": 32,
+  "qk_head_dim": 128,
+  "qk_pos_emb_head_dim": 64,
+  "fp8_format": null,
+  "fp8_recipe": "delayed",
+  "fp8_amax_history_len": 1024,
+  "fp8_amax_compute_algo": "max",
+  "fp8_param_gather": false,
+  "fp16": false,
+  "bf16": true,
+  "apply_query_key_layer_scaling": false,
+  "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": "my_aff",
+  "wandb_exp_name": "Qwen3-4B-Thinking-2507-db-data",
+  "wandb_save_dir": null,
+  "eval_iters": -1,
+  "eval_interval": 1490,
+  "seq_length": 8192,
+  "num_workers": 8,
+  "extra_megatron_kwargs": {},
+  "add_version": true,
+  "rank": 0,
+  "global_world_size": 8,
+  "local_world_size": 8,
+  "model_suffix": "Qwen3-4B-Thinking-2507",
+  "model_info": "ModelInfo(model_type='qwen3_thinking', model_dir='/workspace/Qwen3-4B-Thinking-2507', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, config=None, task_type='causal_lm', num_labels=None)",
+  "model_meta": "ModelMeta(model_type='qwen3_thinking', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-4B-Thinking-2507', hf_model_id='Qwen/Qwen3-4B-Thinking-2507', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-4B-Thinking-2507-FP8', hf_model_id='Qwen/Qwen3-4B-Thinking-2507-FP8', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen3_thinking', get_function=<function get_model_tokenizer_with_flash_attn at 0x789af7853880>, model_arch=None, architectures=['Qwen3ForCausalLM'], additional_saved_files=[], torch_dtype=None, is_multimodal=False, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.51'], tags=[])",
+  "model_dir": "/workspace/Qwen3-4B-Thinking-2507",
+  "hub": "<class 'swift.hub.hub.MSHub'>",
+  "megatron_model_meta": "MegatronModelMeta(megatron_model_type='gpt', model_types=['qwen2', 'qwen2_5', 'qwq', 'qwq_preview', 'qwen2_5_math', 'llama', 'llama3', 'llama3_1', 'llama3_2', 'longwriter_llama3_1', 'codefuse_codellama', 'marco_o1', 'deepseek', 'deepseek_r1_distill', 'yi', 'yi_coder', 'sus', 'skywork_o1', 'openbuddy_llama', 'openbuddy_llama3', 'megrez', 'reflection', 'numina', 'ziya', 'mengzi3', 'qwen3', 'qwen3_thinking', 'qwen3_nothinking', 'qwen2_moe', 'qwen3_moe', 'qwen3_moe_thinking', 'qwen3_coder', 'internlm3', 'mimo', 'mimo_rl', 'moonlight', 'deepseek_moe', 'deepseek_v2', 'deepseek_v2_5', 'deepseek_r1', 'dots1', 'ernie', 'glm4_5', 'deepseek_v3_1', 'ernie_thinking'], convert_mcore2hf=<function convert_mcore2hf at 0x789a608fe2a0>, convert_hf2mcore=<function convert_hf2mcore at 0x789a608fdda0>, model_cls=<class 'swift.megatron.model.gpt_model.GPTModel'>, convert_hf_config=<function convert_gpt_hf_config at 0x789a608c9b20>, get_transformer_layer_spec=None, model_provider=<function model_provider at 0x789a608bf060>, visual_cls=None, extra_args_provider=None)",
+  "extra_args": {
+    "model": "Qwen3-4B-Thinking-2507",
+    "model_type": "qwen3_thinking",
+    "model_revision": null,
+    "task_type": "causal_lm",
+    "torch_dtype": "bfloat16",
+    "attn_impl": null,
+    "new_special_tokens": [],
+    "num_labels": null,
+    "problem_type": null,
+    "rope_scaling": null,
+    "device_map": null,
+    "max_memory": {},
+    "max_model_len": null,
+    "local_repo_path": null,
+    "init_strategy": null,
+    "template": "qwen3_thinking",
+    "system": null,
+    "max_length": 8192,
+    "truncation_strategy": "delete",
+    "max_pixels": null,
+    "agent_template": null,
+    "norm_bbox": null,
+    "use_chat_template": true,
+    "padding_free": true,
+    "padding_side": "right",
+    "sequence_parallel_size": 1,
+    "response_prefix": null,
+    "template_backend": "swift",
+    "dataset": [],
+    "val_dataset": [],
+    "split_dataset_ratio": 0.0,
+    "data_seed": 42,
+    "dataset_num_proc": 8,
+    "load_from_cache_file": false,
+    "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": "/workspace/Qwen3-4B-Thinking-2507-mcore",
+    "lora_modules": [],
+    "tuner_backend": "peft",
+    "train_type": "full",
+    "adapters": [],
+    "external_plugins": [],
+    "model_kwargs": {},
+    "load_args": false,
+    "load_data_args": false,
+    "packing": true,
+    "packing_length": 8192,
+    "lazy_tokenize": false,
+    "cached_dataset": [
+      "my_data_output_only_cached"
+    ],
+    "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,
+    "rlhf_type": null,
+    "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": null,
+    "desirable_weight": 1.0,
+    "undesirable_weight": 1.0,
+    "calculate_KL": null,
+    "center_rewards_coefficient": null,
+    "padded_vocab_size": 151936,
+    "initialize_embedding": false,
+    "mlp_padding_free": false,
+    "dataloader_persistent_workers": true,
+    "dataloader_prefetch_factor": 10,
+    "architectures": "Qwen3ForCausalLM",
+    "llm_architectures": null,
+    "max_epochs": 4,
+    "enable_dft_loss": false,
+    "enable_channel_loss": false,
+    "original_max_position_embeddings": null,
+    "partial_rotary_factor": null,
+    "use_shared_expert_gate": false,
+    "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": false,
+    "add_version": true,
+    "model_info": "ModelInfo(model_type='qwen3_thinking', model_dir='/workspace/Qwen3-4B-Thinking-2507', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, config=None, task_type='causal_lm', num_labels=None)",
+    "model_meta": "ModelMeta(model_type='qwen3_thinking', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-4B-Thinking-2507', hf_model_id='Qwen/Qwen3-4B-Thinking-2507', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-4B-Thinking-2507-FP8', hf_model_id='Qwen/Qwen3-4B-Thinking-2507-FP8', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen3_thinking', get_function=<function get_model_tokenizer_with_flash_attn at 0x789af7853880>, model_arch=None, architectures=['Qwen3ForCausalLM'], additional_saved_files=[], torch_dtype=None, is_multimodal=False, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.51'], tags=[])",
+    "megatron_model_meta": "MegatronModelMeta(megatron_model_type='gpt', model_types=['qwen2', 'qwen2_5', 'qwq', 'qwq_preview', 'qwen2_5_math', 'llama', 'llama3', 'llama3_1', 'llama3_2', 'longwriter_llama3_1', 'codefuse_codellama', 'marco_o1', 'deepseek', 'deepseek_r1_distill', 'yi', 'yi_coder', 'sus', 'skywork_o1', 'openbuddy_llama', 'openbuddy_llama3', 'megrez', 'reflection', 'numina', 'ziya', 'mengzi3', 'qwen3', 'qwen3_thinking', 'qwen3_nothinking', 'qwen2_moe', 'qwen3_moe', 'qwen3_moe_thinking', 'qwen3_coder', 'internlm3', 'mimo', 'mimo_rl', 'moonlight', 'deepseek_moe', 'deepseek_v2', 'deepseek_v2_5', 'deepseek_r1', 'dots1', 'ernie', 'glm4_5', 'deepseek_v3_1', 'ernie_thinking'], convert_mcore2hf=<function convert_mcore2hf at 0x789a608fe2a0>, convert_hf2mcore=<function convert_hf2mcore at 0x789a608fdda0>, model_cls=<class 'swift.megatron.model.gpt_model.GPTModel'>, convert_hf_config=<function convert_gpt_hf_config at 0x789a608c9b20>, get_transformer_layer_spec=None, model_provider=<function model_provider at 0x789a608bf060>, visual_cls=None, extra_args_provider=None)"
+  }
+}