Add training artifacts for v0-20251209-025758

training_artifacts/v0-20251209-025758/args.json

@@ -0,0 +1,353 @@
+{
+  "output_dir": "/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758",
+  "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": 32,
+  "per_device_eval_batch_size": 32,
+  "per_gpu_train_batch_size": null,
+  "per_gpu_eval_batch_size": null,
+  "gradient_accumulation_steps": 1,
+  "eval_accumulation_steps": 1,
+  "eval_delay": 0,
+  "torch_empty_cache_steps": 50,
+  "learning_rate": 0.0002,
+  "weight_decay": 0.1,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.95,
+  "adam_epsilon": 1e-08,
+  "max_grad_norm": 1.0,
+  "num_train_epochs": 2.0,
+  "max_steps": -1,
+  "lr_scheduler_type": "cosine",
+  "lr_scheduler_kwargs": null,
+  "warmup_ratio": 0.1,
+  "warmup_steps": 20,
+  "log_level": "passive",
+  "log_level_replica": "warning",
+  "log_on_each_node": true,
+  "logging_dir": "/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758/runs",
+  "logging_strategy": "steps",
+  "logging_first_step": true,
+  "logging_steps": 5,
+  "logging_nan_inf_filter": true,
+  "save_strategy": "steps",
+  "save_steps": 40.0,
+  "save_total_limit": 3,
+  "save_safetensors": true,
+  "save_on_each_node": false,
+  "save_only_model": false,
+  "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,
+  "bf16": true,
+  "fp16": false,
+  "fp16_opt_level": "O1",
+  "half_precision_backend": "auto",
+  "bf16_full_eval": false,
+  "fp16_full_eval": true,
+  "tf32": null,
+  "local_rank": 0,
+  "ddp_backend": null,
+  "tpu_num_cores": null,
+  "tpu_metrics_debug": false,
+  "debug": null,
+  "dataloader_drop_last": false,
+  "eval_steps": 20.0,
+  "dataloader_num_workers": 8,
+  "dataloader_prefetch_factor": null,
+  "past_index": -1,
+  "run_name": "Qwen3-VL-8B-rslora-r32-2",
+  "disable_tqdm": null,
+  "remove_unused_columns": false,
+  "label_names": null,
+  "load_best_model_at_end": true,
+  "metric_for_best_model": "eval_loss",
+  "greater_is_better": false,
+  "ignore_data_skip": false,
+  "fsdp": null,
+  "fsdp_min_num_params": 0,
+  "fsdp_config": null,
+  "fsdp_transformer_layer_cls_to_wrap": null,
+  "accelerator_config": {
+    "dispatch_batches": false
+  },
+  "parallelism_config": null,
+  "deepspeed": null,
+  "label_smoothing_factor": 0.0,
+  "optim": "adamw_torch",
+  "optim_args": null,
+  "adafactor": false,
+  "group_by_length": false,
+  "length_column_name": "length",
+  "report_to": [
+    "wandb",
+    "tensorboard"
+  ],
+  "project": "quantum-assistant",
+  "trackio_space_id": "trackio",
+  "ddp_find_unused_parameters": null,
+  "ddp_bucket_cap_mb": null,
+  "ddp_broadcast_buffers": null,
+  "dataloader_pin_memory": true,
+  "dataloader_persistent_workers": false,
+  "skip_memory_metrics": true,
+  "use_legacy_prediction_loop": false,
+  "push_to_hub": false,
+  "resume_from_checkpoint": null,
+  "hub_model_id": "samuellimabraz/Qwen3-VL-8B-rslora-r32-2",
+  "hub_strategy": "every_save",
+  "hub_token": null,
+  "hub_private_repo": true,
+  "hub_always_push": false,
+  "hub_revision": null,
+  "gradient_checkpointing": true,
+  "gradient_checkpointing_kwargs": null,
+  "include_inputs_for_metrics": false,
+  "include_for_metrics": [],
+  "eval_do_concat_batches": true,
+  "fp16_backend": "auto",
+  "push_to_hub_model_id": null,
+  "push_to_hub_organization": null,
+  "push_to_hub_token": null,
+  "mp_parameters": "",
+  "auto_find_batch_size": false,
+  "full_determinism": false,
+  "torchdynamo": null,
+  "ray_scope": "last",
+  "ddp_timeout": 18000000,
+  "torch_compile": false,
+  "torch_compile_backend": null,
+  "torch_compile_mode": null,
+  "include_tokens_per_second": false,
+  "include_num_input_tokens_seen": false,
+  "neftune_noise_alpha": null,
+  "optim_target_modules": null,
+  "batch_eval_metrics": false,
+  "eval_on_start": false,
+  "use_liger_kernel": false,
+  "liger_kernel_config": null,
+  "eval_use_gather_object": false,
+  "average_tokens_across_devices": true,
+  "sortish_sampler": false,
+  "predict_with_generate": false,
+  "generation_max_length": null,
+  "generation_num_beams": null,
+  "generation_config": null,
+  "tuner_backend": "peft",
+  "vit_gradient_checkpointing": null,
+  "router_aux_loss_coef": 0.0,
+  "enable_dft_loss": false,
+  "enable_channel_loss": false,
+  "check_model": true,
+  "acc_strategy": "token",
+  "train_dataloader_shuffle": true,
+  "max_epochs": null,
+  "aligner_lr": null,
+  "vit_lr": null,
+  "use_logits_to_keep": null,
+  "ds3_gather_for_generation": true,
+  "resume_only_model": false,
+  "optimizer": null,
+  "loss_type": null,
+  "metric": null,
+  "eval_use_evalscope": false,
+  "eval_dataset": [],
+  "eval_dataset_args": null,
+  "eval_limit": null,
+  "eval_generation_config": null,
+  "extra_eval_args": null,
+  "use_flash_ckpt": false,
+  "use_ray": false,
+  "ray_exp_name": null,
+  "device_groups": null,
+  "model": "Qwen/Qwen3-VL-8B-Instruct",
+  "model_type": "qwen3_vl",
+  "model_revision": null,
+  "task_type": "causal_lm",
+  "torch_dtype": "bfloat16",
+  "attn_impl": "flash_attn",
+  "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_vl",
+  "system": "You are a quantum computing expert assistant specializing in Qiskit.\nProvide accurate, clear, and well-structured responses about quantum computing concepts,\nalgorithms, and code implementation. Use Qiskit 2.0 best practices.\n",
+  "max_length": 262144,
+  "truncation_strategy": "delete",
+  "max_pixels": 1003520,
+  "agent_template": null,
+  "norm_bbox": null,
+  "use_chat_template": true,
+  "padding_free": true,
+  "padding_side": "left",
+  "loss_scale": "default",
+  "sequence_parallel_size": 1,
+  "response_prefix": null,
+  "template_backend": "swift",
+  "dataset": [
+    "./outputs/quantum-assistant/train.jsonl"
+  ],
+  "val_dataset": [
+    "./outputs/quantum-assistant/validation.jsonl"
+  ],
+  "split_dataset_ratio": 0.0,
+  "dataset_num_proc": 8,
+  "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,
+  "model_name": [
+    "Qwen3-VL-8B-rslora-r32-2"
+  ],
+  "model_author": [
+    "samuellimabraz"
+  ],
+  "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": 4096,
+  "temperature": 0.0,
+  "top_k": null,
+  "top_p": null,
+  "repetition_penalty": null,
+  "num_beams": 1,
+  "stream": false,
+  "stop_words": [],
+  "logprobs": false,
+  "top_logprobs": null,
+  "ckpt_dir": null,
+  "lora_modules": [],
+  "train_type": "lora",
+  "adapters": [],
+  "external_plugins": [],
+  "model_kwargs": {},
+  "load_args": false,
+  "load_data_args": false,
+  "packing": false,
+  "packing_length": null,
+  "lazy_tokenize": true,
+  "cached_dataset": [],
+  "custom_register_path": [],
+  "use_hf": true,
+  "ignore_args_error": false,
+  "use_swift_lora": false,
+  "freeze_parameters": [],
+  "freeze_parameters_regex": null,
+  "freeze_parameters_ratio": 0.0,
+  "trainable_parameters": [],
+  "trainable_parameters_regex": null,
+  "freeze_llm": false,
+  "freeze_vit": true,
+  "freeze_aligner": false,
+  "target_modules": [
+    "all-linear"
+  ],
+  "target_regex": null,
+  "target_parameters": null,
+  "modules_to_save": [],
+  "lora_rank": 32,
+  "lora_alpha": 64,
+  "lora_dropout": 0.15,
+  "lora_bias": "none",
+  "lora_dtype": null,
+  "lorap_lr_ratio": null,
+  "use_rslora": true,
+  "use_dora": false,
+  "lora_ga_batch_size": 2,
+  "lora_ga_iters": 2,
+  "lora_ga_max_length": 1024,
+  "lora_ga_direction": "ArB2r",
+  "lora_ga_scale": "stable",
+  "lora_ga_stable_gamma": 16,
+  "init_weights": true,
+  "fourier_n_frequency": 2000,
+  "fourier_scaling": 300.0,
+  "boft_block_size": 4,
+  "boft_block_num": 0,
+  "boft_n_butterfly_factor": 1,
+  "boft_dropout": 0.0,
+  "vera_rank": 256,
+  "vera_projection_prng_key": 0,
+  "vera_dropout": 0.0,
+  "vera_d_initial": 0.1,
+  "adapter_act": "gelu",
+  "adapter_length": 128,
+  "use_galore": false,
+  "galore_target_modules": null,
+  "galore_rank": 128,
+  "galore_update_proj_gap": 50,
+  "galore_scale": 1.0,
+  "galore_proj_type": "std",
+  "galore_optim_per_parameter": false,
+  "galore_with_embedding": false,
+  "galore_quantization": false,
+  "galore_proj_quant": false,
+  "galore_proj_bits": 4,
+  "galore_proj_group_size": 256,
+  "galore_cos_threshold": 0.4,
+  "galore_gamma_proj": 2,
+  "galore_queue_size": 5,
+  "adalora_target_r": 8,
+  "adalora_init_r": 12,
+  "adalora_tinit": 0,
+  "adalora_tfinal": 0,
+  "adalora_deltaT": 1,
+  "adalora_beta1": 0.85,
+  "adalora_beta2": 0.85,
+  "adalora_orth_reg_weight": 0.5,
+  "llamapro_num_new_blocks": 4,
+  "llamapro_num_groups": null,
+  "lisa_activated_layers": 0,
+  "lisa_step_interval": 20,
+  "reft_layer_key": null,
+  "reft_layers": null,
+  "reft_rank": 4,
+  "reft_intervention_type": "LoreftIntervention",
+  "reft_args": null,
+  "swanlab_token": null,
+  "swanlab_project": null,
+  "swanlab_workspace": null,
+  "swanlab_exp_name": null,
+  "swanlab_lark_webhook_url": null,
+  "swanlab_lark_secret": null,
+  "swanlab_mode": "cloud",
+  "add_version": true,
+  "create_checkpoint_symlink": false,
+  "zero_hpz_partition_size": null,
+  "deepspeed_autotp_size": null,
+  "early_stop_interval": null,
+  "rank": 0,
+  "global_world_size": 1,
+  "local_world_size": 1,
+  "model_suffix": "Qwen3-VL-8B-Instruct",
+  "model_info": "ModelInfo(model_type='qwen3_vl', model_dir='/workspace/.cache/huggingface/hub/models--Qwen--Qwen3-VL-8B-Instruct/snapshots/0c351dd01ed87e9c1b53cbc748cba10e6187ff3b', 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=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct', hf_model_id='Qwen/Qwen3-VL-2B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking', hf_model_id='Qwen/Qwen3-VL-2B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-2B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct', hf_model_id='Qwen/Qwen3-VL-4B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking', hf_model_id='Qwen/Qwen3-VL-4B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-4B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct', hf_model_id='Qwen/Qwen3-VL-8B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking', hf_model_id='Qwen/Qwen3-VL-8B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-8B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct', hf_model_id='Qwen/Qwen3-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking', hf_model_id='Qwen/Qwen3-VL-32B-Thinking', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Instruct-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', hf_model_id='Qwen/Qwen3-VL-32B-Thinking-FP8', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen3_vl', get_function=<function get_model_tokenizer_qwen3_vl at 0x7963956dfb00>, 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', 'lm_head'], 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, is_reranker=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.57', 'qwen_vl_utils>=0.0.14', 'decord'], tags=['vision', 'video'])",
+  "model_dir": "/workspace/.cache/huggingface/hub/models--Qwen--Qwen3-VL-8B-Instruct/snapshots/0c351dd01ed87e9c1b53cbc748cba10e6187ff3b",
+  "hub": "<class 'swift.hub.hub.HFHub'>",
+  "evaluation_strategy": "steps",
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, prediction_loss_only=False, per_device_train_batch_size=32, per_device_eval_batch_size=32, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=1, eval_accumulation_steps=1, eval_delay=0, torch_empty_cache_steps=50, learning_rate=0.0002, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=2.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.1, warmup_steps=20, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758/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=40, save_total_limit=3, save_safetensors=True, save_on_each_node=False, save_only_model=False, 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, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=True, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=20, dataloader_num_workers=8, dataloader_prefetch_factor=10, past_index=-1, run_name='Qwen3-VL-8B-rslora-r32-2', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=True, metric_for_best_model='eval_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}, 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), parallelism_config=None, deepspeed=None, 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', 'tensorboard'], project='quantum-assistant', trackio_space_id='trackio', 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='samuellimabraz/Qwen3-VL-8B-rslora-r32-2', hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=True, hub_always_push=False, hub_revision=None, 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=18000000, 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, liger_kernel_config=None, 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, tuner_backend='peft', vit_gradient_checkpointing=True, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, max_epochs=None, aligner_lr=None, vit_lr=None, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer=None, loss_type=None, metric=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, use_flash_ckpt=False, sft_alpha=0, chord_sft_dataset=[], chord_sft_per_device_train_batch_size=None, chord_enable_phi_function=False, chord_mu_warmup_steps=None, chord_mu_decay_steps=None, chord_mu_peak=None, chord_mu_valley=None, train_type='lora', local_repo_path=None, galore_config=None, padding_side='left', padding_free=True, task_type='causal_lm', problem_type=None)"
+}

training_artifacts/v0-20251209-025758/base.yaml

@@ -0,0 +1,122 @@
+# =============================================================================
+# Base Configuration for Quantum Assistant Fine-tuning
+# All experiments inherit from this file
+# =============================================================================
+
+# -----------------------------------------------------------------------------
+# Model Configuration
+# -----------------------------------------------------------------------------
+model: Qwen/Qwen3-VL-8B-Instruct
+model_author: samuellimabraz
+
+# System Prompt (consistent with evaluation)
+system: |
+  You are a quantum computing expert assistant specializing in Qiskit.
+  Provide accurate, clear, and well-structured responses about quantum computing concepts,
+  algorithms, and code implementation. Use Qiskit 2.0 best practices.
+
+# -----------------------------------------------------------------------------
+# Dataset Configuration
+# -----------------------------------------------------------------------------
+dataset:
+  - ./outputs/quantum-assistant/train.jsonl
+val_dataset:
+  - ./outputs/quantum-assistant/validation.jsonl
+load_from_cache_file: true
+data_seed: 42
+
+# -----------------------------------------------------------------------------
+# Model Precision & Attention
+# -----------------------------------------------------------------------------
+torch_dtype: bfloat16
+bf16: true
+fp16: false
+attn_impl: flash_attn
+
+# -----------------------------------------------------------------------------
+# Tokenization & Sequence Settings
+# -----------------------------------------------------------------------------
+padding_side: left
+padding_free: true
+lazy_tokenize: true
+max_new_tokens: 4096
+max_pixels: 1003520
+
+# -----------------------------------------------------------------------------
+# LoRA Configuration
+# -----------------------------------------------------------------------------
+train_type: lora
+lora_dropout: 0.15
+target_modules: all-linear 
+
+# Component-specific learning rates (optional, uncomment to use)
+# aligner_lr: 1e-4     # Lower than main LR for stable training
+# vit_lr: 1e-5         # Only if training ViT (not recommended)
+
+# -----------------------------------------------------------------------------
+# Training Configuration
+# -----------------------------------------------------------------------------
+# Optimization
+learning_rate: 2e-4
+lr_scheduler_type: cosine
+optim: adamw_torch
+weight_decay: 0.1
+
+# Batch & Accumulation
+per_device_train_batch_size: 32
+gradient_accumulation_steps: 1
+torch_empty_cache_steps: 50
+
+# Epochs & Steps
+num_train_epochs: 2
+warmup_steps: 20
+warmup_ratio: 0.1
+
+# Memory Optimization
+gradient_checkpointing: true
+
+# -----------------------------------------------------------------------------
+# Evaluation Configuration
+# -----------------------------------------------------------------------------
+per_device_eval_batch_size: 32
+eval_accumulation_steps: 1
+eval_strategy: steps
+eval_steps: 20
+metric_for_best_model: eval_loss
+load_best_model_at_end: true
+fp16_full_eval: true
+
+# -----------------------------------------------------------------------------
+# Saving Configuration
+# -----------------------------------------------------------------------------
+save_strategy: steps
+save_steps: 40
+save_total_limit: 3
+
+# -----------------------------------------------------------------------------
+# Logging Configuration
+# -----------------------------------------------------------------------------
+logging_first_step: true
+logging_steps: 5
+logging_strategy: steps
+report_to:
+  - wandb
+  - tensorboard
+
+# Project naming
+project: quantum-assistant
+
+# -----------------------------------------------------------------------------
+# Data Loading
+# -----------------------------------------------------------------------------
+dataloader_num_workers: 8
+dataset_num_proc: 8
+remove_unused_columns: false
+
+# -----------------------------------------------------------------------------
+# HuggingFace Hub Configuration
+# -----------------------------------------------------------------------------
+use_hf: true
+push_to_hub: false
+hub_private_repo: true
+

training_artifacts/v0-20251209-025758/experiments.yaml

@@ -0,0 +1,164 @@
+# =============================================================================
+# PEFT Experiments Configuration
+# Each experiment overrides base.yaml with specific parameters
+# 
+# Usage: ./scripts/experiment.sh <experiment_name>
+#   Example: ./scripts/experiment.sh pissa
+# =============================================================================
+
+# =============================================================================
+# Initial Experiments: Compare PEFT methods (1 epoch, r=16)
+# =============================================================================
+
+lora:
+  model_name: Qwen3-VL-8B-lora
+  output_dir: ./outputs/train/exp-lora
+  run_name: Qwen3-VL-8B-lora
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-lora
+  # LoRA settings
+  init_weights: "true"
+  use_rslora: false
+  use_dora: false
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# Baseline: rsLoRA with standard initialization
+rslora:
+  model_name: Qwen3-VL-8B-rslora
+  output_dir: ./outputs/train/exp-rslora
+  run_name: Qwen3-VL-8B-rslora
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-rslora
+  # LoRA settings
+  init_weights: "true"
+  use_rslora: true
+  use_dora: false
+  # Freeze settings
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# PiSSA: SVD-based initialization (faster convergence)
+pissa:
+  model_name: Qwen3-VL-8B-pissa
+  output_dir: ./outputs/train/exp-pissa
+  run_name: Qwen3-VL-8B-pissa
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-pissa
+  # LoRA settings
+  init_weights: "pissa"
+  use_rslora: true
+  use_dora: false
+  # Freeze settings
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# OLoRA: QR orthonormal initialization (stable training)
+olora:
+  model_name: Qwen3-VL-8B-olora
+  output_dir: ./outputs/train/exp-olora
+  run_name: Qwen3-VL-8B-olora
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-olora
+  # LoRA settings
+  init_weights: "olora"
+  use_rslora: true
+  use_dora: false
+  # Freeze settings
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# DoRA: magnitude-direction decomposition (best quality)
+dora:
+  model_name: Qwen3-VL-8B-dora
+  output_dir: ./outputs/train/exp-dora
+  run_name: Qwen3-VL-8B-dora
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-dora
+  # LoRA settings
+  init_weights: "true"
+  use_rslora: true
+  use_dora: true
+  # Freeze settings
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# =============================================================================
+# Ablation Experiments
+# =============================================================================
+
+# Ablation: freeze aligner (compare multimodal learning)
+rslora_frozen_aligner:
+  model_name: Qwen3-VL-8B-rslora-frozen
+  output_dir: ./outputs/train/exp-rslora-frozen
+  run_name: Qwen3-VL-8B-rslora-frozen
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-rslora-frozen
+  # LoRA settings
+  init_weights: "true"
+  use_rslora: true
+  use_dora: false
+  # Freeze settings (aligner frozen for comparison)
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: true
+
+rslora_r32:
+  model_name: Qwen3-VL-8B-rslora-r32-2
+  output_dir: ./outputs/train/exp-rslora-r32-2
+  run_name: Qwen3-VL-8B-rslora-r32-2
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-rslora-r32-2
+  lora_rank: 32
+  lora_alpha: 64  
+  init_weights: "true"
+  use_rslora: true
+  use_dora: false
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+
+rslora_r64:
+  model_name: Qwen3-VL-8B-rslora-r64
+  output_dir: ./outputs/train/exp-rslora-r64
+  run_name: Qwen3-VL-8B-rslora-r64
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-rslora-r64
+  lora_rank: 64
+  lora_alpha: 128  
+  init_weights: "true"
+  use_rslora: true
+  use_dora: false
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# =============================================================================
+# Final Experiments (Best Result Search)
+# =============================================================================
+
+# High Rank rsLoRA with more epochs
+rslora_r128:
+  model_name: Qwen3-VL-8B-rslora-r128
+  output_dir: ./outputs/train/exp-rslora-r128
+  run_name: Qwen3-VL-8B-rslora-r128
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-rslora-r128
+  lora_rank: 128
+  lora_alpha: 256
+  init_weights: "true"
+  use_rslora: true
+  use_dora: false
+  freeze_llm: false
+  freeze_vit: true
+  freeze_aligner: false
+
+# Full Fine-Tuning (Upper bound for quality, requires more VRAM)
+full_ft:
+  model_name: Qwen3-VL-8B-full
+  output_dir: ./outputs/train/exp-full
+  run_name: Qwen3-VL-8B-full
+  hub_model_id: samuellimabraz/Qwen3-VL-8B-full
+  train_type: full
+  learning_rate: 1e-5  # Lower LR for full finetuning
+  num_train_epochs: 3
+  freeze_llm: false
+  freeze_vit: true     # Protect vision encoder
+  freeze_aligner: false

training_artifacts/v0-20251209-025758/logging.jsonl

@@ -0,0 +1,95 @@
+{"loss": 1.27906477, "grad_norm": 7.51647711, "learning_rate": 1e-05, "token_acc": 0.78187566, "epoch": 0.00546448, "global_step/max_steps": "1/366", "percentage": "0.27%", "elapsed_time": "5s", "remaining_time": "32m 7s", "memory(GiB)": 38.73, "train_speed(iter/s)": 0.189367}
+{"loss": 1.14319742, "grad_norm": 2.46747923, "learning_rate": 5e-05, "token_acc": 0.77237399, "epoch": 0.0273224, "global_step/max_steps": "5/366", "percentage": "1.37%", "elapsed_time": "20s", "remaining_time": "24m 5s", "memory(GiB)": 54.46, "train_speed(iter/s)": 0.24982}
+{"loss": 0.92578173, "grad_norm": 1.4030453, "learning_rate": 0.0001, "token_acc": 0.7604315, "epoch": 0.05464481, "global_step/max_steps": "10/366", "percentage": "2.73%", "elapsed_time": "37s", "remaining_time": "22m 26s", "memory(GiB)": 54.46, "train_speed(iter/s)": 0.264305}
+{"loss": 0.85726051, "grad_norm": 1.33702552, "learning_rate": 0.00015, "token_acc": 0.76674365, "epoch": 0.08196721, "global_step/max_steps": "15/366", "percentage": "4.10%", "elapsed_time": "57s", "remaining_time": "22m 30s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.259843}
+{"loss": 0.62358894, "grad_norm": 1.47015607, "learning_rate": 0.0002, "token_acc": 0.8272145, "epoch": 0.10928962, "global_step/max_steps": "20/366", "percentage": "5.46%", "elapsed_time": "1m 15s", "remaining_time": "21m 46s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.264894}
+{"eval_loss": 0.73444343, "eval_runtime": 47.1079, "eval_samples_per_second": 26.301, "eval_steps_per_second": 0.828, "eval_token_acc": 0.79498505, "epoch": 0.10928962, "global_step/max_steps": "20/366", "percentage": "5.46%", "elapsed_time": "2m 2s", "remaining_time": "35m 21s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.163115}
+{"loss": 0.67104912, "grad_norm": 0.88824362, "learning_rate": 0.0001999, "token_acc": 0.81274377, "epoch": 0.13661202, "global_step/max_steps": "25/366", "percentage": "6.83%", "elapsed_time": "2m 21s", "remaining_time": "32m 3s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.177277}
+{"loss": 0.68390784, "grad_norm": 1.14266908, "learning_rate": 0.00019959, "token_acc": 0.80918728, "epoch": 0.16393443, "global_step/max_steps": "30/366", "percentage": "8.20%", "elapsed_time": "2m 39s", "remaining_time": "29m 42s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.188456}
+{"loss": 0.63831177, "grad_norm": 0.83876228, "learning_rate": 0.00019907, "token_acc": 0.82558862, "epoch": 0.19125683, "global_step/max_steps": "35/366", "percentage": "9.56%", "elapsed_time": "2m 57s", "remaining_time": "28m 0s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.196926}
+{"loss": 0.76150837, "grad_norm": 1.06894028, "learning_rate": 0.00019836, "token_acc": 0.78989403, "epoch": 0.21857923, "global_step/max_steps": "40/366", "percentage": "10.93%", "elapsed_time": "3m 16s", "remaining_time": "26m 43s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.203327}
+{"eval_loss": 0.69627213, "eval_runtime": 46.9273, "eval_samples_per_second": 26.403, "eval_steps_per_second": 0.831, "eval_token_acc": 0.8038186, "epoch": 0.21857923, "global_step/max_steps": "40/366", "percentage": "10.93%", "elapsed_time": "4m 3s", "remaining_time": "33m 5s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.164165}
+{"loss": 0.65252018, "grad_norm": 0.8042981, "learning_rate": 0.00019743, "token_acc": 0.81617397, "epoch": 0.24590164, "global_step/max_steps": "45/366", "percentage": "12.30%", "elapsed_time": "4m 23s", "remaining_time": "31m 21s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.170626}
+{"loss": 0.66020288, "grad_norm": 1.05325747, "learning_rate": 0.00019631, "token_acc": 0.81153249, "epoch": 0.27322404, "global_step/max_steps": "50/366", "percentage": "13.66%", "elapsed_time": "4m 41s", "remaining_time": "29m 41s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.177425}
+{"loss": 0.68664289, "grad_norm": 0.98790455, "learning_rate": 0.00019499, "token_acc": 0.8062056, "epoch": 0.30054645, "global_step/max_steps": "55/366", "percentage": "15.03%", "elapsed_time": "5m 0s", "remaining_time": "28m 20s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.182933}
+{"loss": 0.67267942, "grad_norm": 0.89244473, "learning_rate": 0.00019348, "token_acc": 0.81076363, "epoch": 0.32786885, "global_step/max_steps": "60/366", "percentage": "16.39%", "elapsed_time": "5m 20s", "remaining_time": "27m 14s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.187263}
+{"eval_loss": 0.67779845, "eval_runtime": 47.0363, "eval_samples_per_second": 26.341, "eval_steps_per_second": 0.829, "eval_token_acc": 0.80857819, "epoch": 0.32786885, "global_step/max_steps": "60/366", "percentage": "16.39%", "elapsed_time": "6m 7s", "remaining_time": "31m 13s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.16329}
+{"loss": 0.66991453, "grad_norm": 0.947658, "learning_rate": 0.00019177, "token_acc": 0.81701559, "epoch": 0.35519126, "global_step/max_steps": "65/366", "percentage": "17.76%", "elapsed_time": "6m 25s", "remaining_time": "29m 43s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.168726}
+{"loss": 0.66196041, "grad_norm": 0.72777605, "learning_rate": 0.00018987, "token_acc": 0.81384481, "epoch": 0.38251366, "global_step/max_steps": "70/366", "percentage": "19.13%", "elapsed_time": "6m 46s", "remaining_time": "28m 37s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.172316}
+{"loss": 0.64290462, "grad_norm": 0.85720301, "learning_rate": 0.00018779, "token_acc": 0.81829405, "epoch": 0.40983607, "global_step/max_steps": "75/366", "percentage": "20.49%", "elapsed_time": "7m 5s", "remaining_time": "27m 30s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.176304}
+{"loss": 0.69418397, "grad_norm": 0.82464141, "learning_rate": 0.00018552, "token_acc": 0.80308035, "epoch": 0.43715847, "global_step/max_steps": "80/366", "percentage": "21.86%", "elapsed_time": "7m 24s", "remaining_time": "26m 27s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.180164}
+{"eval_loss": 0.65976292, "eval_runtime": 46.8466, "eval_samples_per_second": 26.448, "eval_steps_per_second": 0.833, "eval_token_acc": 0.81260744, "epoch": 0.43715847, "global_step/max_steps": "80/366", "percentage": "21.86%", "elapsed_time": "8m 10s", "remaining_time": "29m 14s", "memory(GiB)": 54.47, "train_speed(iter/s)": 0.16297}
+{"loss": 0.72211752, "grad_norm": 0.84521145, "learning_rate": 0.00018308, "token_acc": 0.79385564, "epoch": 0.46448087, "global_step/max_steps": "85/366", "percentage": "23.22%", "elapsed_time": "8m 32s", "remaining_time": "28m 13s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165891}
+{"loss": 0.60683317, "grad_norm": 0.87995595, "learning_rate": 0.00018047, "token_acc": 0.83045028, "epoch": 0.49180328, "global_step/max_steps": "90/366", "percentage": "24.59%", "elapsed_time": "8m 49s", "remaining_time": "27m 3s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169998}
+{"loss": 0.66692858, "grad_norm": 0.80485332, "learning_rate": 0.0001777, "token_acc": 0.81494842, "epoch": 0.51912568, "global_step/max_steps": "95/366", "percentage": "25.96%", "elapsed_time": "9m 8s", "remaining_time": "26m 5s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.173063}
+{"loss": 0.61119962, "grad_norm": 0.87856847, "learning_rate": 0.00017476, "token_acc": 0.82720454, "epoch": 0.54644809, "global_step/max_steps": "100/366", "percentage": "27.32%", "elapsed_time": "9m 27s", "remaining_time": "25m 8s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.176321}
+{"eval_loss": 0.6528914, "eval_runtime": 46.5122, "eval_samples_per_second": 26.638, "eval_steps_per_second": 0.838, "eval_token_acc": 0.81378491, "epoch": 0.54644809, "global_step/max_steps": "100/366", "percentage": "27.32%", "elapsed_time": "10m 13s", "remaining_time": "27m 12s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162956}
+{"loss": 0.64105968, "grad_norm": 0.88768315, "learning_rate": 0.00017167, "token_acc": 0.82062158, "epoch": 0.57377049, "global_step/max_steps": "105/366", "percentage": "28.69%", "elapsed_time": "10m 32s", "remaining_time": "26m 12s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165953}
+{"loss": 0.61745596, "grad_norm": 0.78273153, "learning_rate": 0.00016843, "token_acc": 0.82540293, "epoch": 0.6010929, "global_step/max_steps": "110/366", "percentage": "30.05%", "elapsed_time": "10m 50s", "remaining_time": "25m 14s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169065}
+{"loss": 0.63137207, "grad_norm": 0.75863063, "learning_rate": 0.00016505, "token_acc": 0.81579435, "epoch": 0.6284153, "global_step/max_steps": "115/366", "percentage": "31.42%", "elapsed_time": "11m 10s", "remaining_time": "24m 22s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.171618}
+{"loss": 0.68505793, "grad_norm": 0.87702191, "learning_rate": 0.00016153, "token_acc": 0.81127324, "epoch": 0.6557377, "global_step/max_steps": "120/366", "percentage": "32.79%", "elapsed_time": "11m 28s", "remaining_time": "23m 31s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.174222}
+{"eval_loss": 0.64290112, "eval_runtime": 46.8791, "eval_samples_per_second": 26.43, "eval_steps_per_second": 0.832, "eval_token_acc": 0.81582687, "epoch": 0.6557377, "global_step/max_steps": "120/366", "percentage": "32.79%", "elapsed_time": "12m 15s", "remaining_time": "25m 8s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163119}
+{"loss": 0.58722191, "grad_norm": 0.77375346, "learning_rate": 0.00015789, "token_acc": 0.83134928, "epoch": 0.68306011, "global_step/max_steps": "125/366", "percentage": "34.15%", "elapsed_time": "12m 36s", "remaining_time": "24m 18s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165228}
+{"loss": 0.60854549, "grad_norm": 1.00976026, "learning_rate": 0.00015413, "token_acc": 0.8255247, "epoch": 0.71038251, "global_step/max_steps": "130/366", "percentage": "35.52%", "elapsed_time": "12m 54s", "remaining_time": "23m 26s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.167781}
+{"loss": 0.52396584, "grad_norm": 0.75767154, "learning_rate": 0.00015026, "token_acc": 0.85100564, "epoch": 0.73770492, "global_step/max_steps": "135/366", "percentage": "36.89%", "elapsed_time": "13m 15s", "remaining_time": "22m 40s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169808}
+{"loss": 0.61536837, "grad_norm": 0.85596806, "learning_rate": 0.00014629, "token_acc": 0.82246136, "epoch": 0.76502732, "global_step/max_steps": "140/366", "percentage": "38.25%", "elapsed_time": "13m 32s", "remaining_time": "21m 51s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.172338}
+{"eval_loss": 0.6382603, "eval_runtime": 46.6167, "eval_samples_per_second": 26.578, "eval_steps_per_second": 0.837, "eval_token_acc": 0.8174018, "epoch": 0.76502732, "global_step/max_steps": "140/366", "percentage": "38.25%", "elapsed_time": "14m 18s", "remaining_time": "23m 6s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162984}
+{"loss": 0.54785199, "grad_norm": 0.86010516, "learning_rate": 0.00014222, "token_acc": 0.83810349, "epoch": 0.79234973, "global_step/max_steps": "145/366", "percentage": "39.62%", "elapsed_time": "14m 37s", "remaining_time": "22m 18s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165161}
+{"loss": 0.6614882, "grad_norm": 0.78265226, "learning_rate": 0.00013806, "token_acc": 0.81246694, "epoch": 0.81967213, "global_step/max_steps": "150/366", "percentage": "40.98%", "elapsed_time": "14m 57s", "remaining_time": "21m 31s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.16722}
+{"loss": 0.61059246, "grad_norm": 0.92743623, "learning_rate": 0.00013382, "token_acc": 0.82696586, "epoch": 0.84699454, "global_step/max_steps": "155/366", "percentage": "42.35%", "elapsed_time": "15m 16s", "remaining_time": "20m 47s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169175}
+{"loss": 0.6324017, "grad_norm": 0.92266405, "learning_rate": 0.00012952, "token_acc": 0.82441923, "epoch": 0.87431694, "global_step/max_steps": "160/366", "percentage": "43.72%", "elapsed_time": "15m 33s", "remaining_time": "20m 1s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.171409}
+{"eval_loss": 0.62636167, "eval_runtime": 46.5309, "eval_samples_per_second": 26.627, "eval_steps_per_second": 0.838, "eval_token_acc": 0.82039766, "epoch": 0.87431694, "global_step/max_steps": "160/366", "percentage": "43.72%", "elapsed_time": "16m 19s", "remaining_time": "21m 1s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.16327}
+{"loss": 0.59900393, "grad_norm": 0.83984584, "learning_rate": 0.00012515, "token_acc": 0.82582212, "epoch": 0.90163934, "global_step/max_steps": "165/366", "percentage": "45.08%", "elapsed_time": "16m 39s", "remaining_time": "20m 17s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165063}
+{"loss": 0.62734685, "grad_norm": 0.88927889, "learning_rate": 0.00012073, "token_acc": 0.82042032, "epoch": 0.92896175, "global_step/max_steps": "170/366", "percentage": "46.45%", "elapsed_time": "16m 57s", "remaining_time": "19m 33s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.167086}
+{"loss": 0.59012399, "grad_norm": 0.7530756, "learning_rate": 0.00011627, "token_acc": 0.83108311, "epoch": 0.95628415, "global_step/max_steps": "175/366", "percentage": "47.81%", "elapsed_time": "17m 17s", "remaining_time": "18m 52s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.168705}
+{"loss": 0.59760904, "grad_norm": 0.9162634, "learning_rate": 0.00011178, "token_acc": 0.82954383, "epoch": 0.98360656, "global_step/max_steps": "180/366", "percentage": "49.18%", "elapsed_time": "17m 36s", "remaining_time": "18m 11s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.170341}
+{"eval_loss": 0.61867183, "eval_runtime": 47.2246, "eval_samples_per_second": 26.236, "eval_steps_per_second": 0.826, "eval_token_acc": 0.82187323, "epoch": 0.98360656, "global_step/max_steps": "180/366", "percentage": "49.18%", "elapsed_time": "18m 23s", "remaining_time": "19m 0s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163053}
+{"loss": 0.54804645, "grad_norm": 0.79565996, "learning_rate": 0.00010726, "token_acc": 0.84721807, "epoch": 1.01092896, "global_step/max_steps": "185/366", "percentage": "50.55%", "elapsed_time": "18m 42s", "remaining_time": "18m 17s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164865}
+{"loss": 0.41285219, "grad_norm": 0.85040879, "learning_rate": 0.00010272, "token_acc": 0.87541022, "epoch": 1.03825137, "global_step/max_steps": "190/366", "percentage": "51.91%", "elapsed_time": "19m 0s", "remaining_time": "17m 36s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166521}
+{"loss": 0.35370812, "grad_norm": 0.91943258, "learning_rate": 9.818e-05, "token_acc": 0.89632551, "epoch": 1.06557377, "global_step/max_steps": "195/366", "percentage": "53.28%", "elapsed_time": "19m 19s", "remaining_time": "16m 56s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.168174}
+{"loss": 0.33952971, "grad_norm": 0.96197706, "learning_rate": 9.365e-05, "token_acc": 0.89531014, "epoch": 1.09289617, "global_step/max_steps": "200/366", "percentage": "54.64%", "elapsed_time": "19m 38s", "remaining_time": "16m 17s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169748}
+{"eval_loss": 0.65192837, "eval_runtime": 46.9396, "eval_samples_per_second": 26.396, "eval_steps_per_second": 0.831, "eval_token_acc": 0.82159004, "epoch": 1.09289617, "global_step/max_steps": "200/366", "percentage": "54.64%", "elapsed_time": "20m 25s", "remaining_time": "16m 56s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163244}
+{"loss": 0.35062573, "grad_norm": 0.85122454, "learning_rate": 8.913e-05, "token_acc": 0.88835808, "epoch": 1.12021858, "global_step/max_steps": "205/366", "percentage": "56.01%", "elapsed_time": "20m 46s", "remaining_time": "16m 18s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164515}
+{"loss": 0.33119357, "grad_norm": 0.84224945, "learning_rate": 8.463e-05, "token_acc": 0.89771326, "epoch": 1.14754098, "global_step/max_steps": "210/366", "percentage": "57.38%", "elapsed_time": "21m 5s", "remaining_time": "15m 40s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165885}
+{"loss": 0.30769742, "grad_norm": 1.07783222, "learning_rate": 8.016e-05, "token_acc": 0.90265816, "epoch": 1.17486339, "global_step/max_steps": "215/366", "percentage": "58.74%", "elapsed_time": "21m 24s", "remaining_time": "15m 2s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.167367}
+{"loss": 0.34732604, "grad_norm": 1.00466919, "learning_rate": 7.573e-05, "token_acc": 0.88834485, "epoch": 1.20218579, "global_step/max_steps": "220/366", "percentage": "60.11%", "elapsed_time": "21m 41s", "remaining_time": "14m 23s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.169016}
+{"eval_loss": 0.65182048, "eval_runtime": 46.6705, "eval_samples_per_second": 26.548, "eval_steps_per_second": 0.836, "eval_token_acc": 0.82160494, "epoch": 1.20218579, "global_step/max_steps": "220/366", "percentage": "60.11%", "elapsed_time": "22m 28s", "remaining_time": "14m 54s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163165}
+{"loss": 0.29864633, "grad_norm": 0.71135694, "learning_rate": 7.135e-05, "token_acc": 0.91037275, "epoch": 1.2295082, "global_step/max_steps": "225/366", "percentage": "61.48%", "elapsed_time": "22m 47s", "remaining_time": "14m 17s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164484}
+{"loss": 0.3305074, "grad_norm": 0.9444766, "learning_rate": 6.703e-05, "token_acc": 0.89911533, "epoch": 1.2568306, "global_step/max_steps": "230/366", "percentage": "62.84%", "elapsed_time": "23m 6s", "remaining_time": "13m 40s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165846}
+{"loss": 0.36874344, "grad_norm": 0.98121345, "learning_rate": 6.278e-05, "token_acc": 0.88416934, "epoch": 1.28415301, "global_step/max_steps": "235/366", "percentage": "64.21%", "elapsed_time": "23m 27s", "remaining_time": "13m 4s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166944}
+{"loss": 0.33786273, "grad_norm": 1.19926894, "learning_rate": 5.861e-05, "token_acc": 0.89503071, "epoch": 1.31147541, "global_step/max_steps": "240/366", "percentage": "65.57%", "elapsed_time": "23m 46s", "remaining_time": "12m 29s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.168199}
+{"eval_loss": 0.65221, "eval_runtime": 46.8505, "eval_samples_per_second": 26.446, "eval_steps_per_second": 0.832, "eval_token_acc": 0.82194279, "epoch": 1.31147541, "global_step/max_steps": "240/366", "percentage": "65.57%", "elapsed_time": "24m 33s", "remaining_time": "12m 53s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162851}
+{"loss": 0.35985045, "grad_norm": 1.10702384, "learning_rate": 5.452e-05, "token_acc": 0.88843258, "epoch": 1.33879781, "global_step/max_steps": "245/366", "percentage": "66.94%", "elapsed_time": "24m 52s", "remaining_time": "12m 16s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164187}
+{"loss": 0.3397682, "grad_norm": 0.98656303, "learning_rate": 5.053e-05, "token_acc": 0.89518264, "epoch": 1.36612022, "global_step/max_steps": "250/366", "percentage": "68.31%", "elapsed_time": "25m 11s", "remaining_time": "11m 41s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165452}
+{"loss": 0.34928138, "grad_norm": 0.99606901, "learning_rate": 4.663e-05, "token_acc": 0.89389346, "epoch": 1.39344262, "global_step/max_steps": "255/366", "percentage": "69.67%", "elapsed_time": "25m 31s", "remaining_time": "11m 6s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166458}
+{"loss": 0.32947588, "grad_norm": 1.01765549, "learning_rate": 4.285e-05, "token_acc": 0.89728272, "epoch": 1.42076503, "global_step/max_steps": "260/366", "percentage": "71.04%", "elapsed_time": "25m 50s", "remaining_time": "10m 32s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.167685}
+{"eval_loss": 0.64265782, "eval_runtime": 46.6918, "eval_samples_per_second": 26.536, "eval_steps_per_second": 0.835, "eval_token_acc": 0.82341836, "epoch": 1.42076503, "global_step/max_steps": "260/366", "percentage": "71.04%", "elapsed_time": "26m 37s", "remaining_time": "10m 51s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162783}
+{"loss": 0.35432878, "grad_norm": 1.00326073, "learning_rate": 3.918e-05, "token_acc": 0.88904529, "epoch": 1.44808743, "global_step/max_steps": "265/366", "percentage": "72.40%", "elapsed_time": "26m 56s", "remaining_time": "10m 16s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163885}
+{"loss": 0.36121664, "grad_norm": 0.87822711, "learning_rate": 3.564e-05, "token_acc": 0.8886253, "epoch": 1.47540984, "global_step/max_steps": "270/366", "percentage": "73.77%", "elapsed_time": "27m 15s", "remaining_time": "9m 41s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.16512}
+{"loss": 0.27267995, "grad_norm": 0.7909438, "learning_rate": 3.224e-05, "token_acc": 0.91560461, "epoch": 1.50273224, "global_step/max_steps": "275/366", "percentage": "75.14%", "elapsed_time": "27m 32s", "remaining_time": "9m 6s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166414}
+{"loss": 0.34633608, "grad_norm": 0.8114062, "learning_rate": 2.897e-05, "token_acc": 0.89605181, "epoch": 1.53005464, "global_step/max_steps": "280/366", "percentage": "76.50%", "elapsed_time": "27m 51s", "remaining_time": "8m 33s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.167522}
+{"eval_loss": 0.63894624, "eval_runtime": 46.8001, "eval_samples_per_second": 26.474, "eval_steps_per_second": 0.833, "eval_token_acc": 0.82422321, "epoch": 1.53005464, "global_step/max_steps": "280/366", "percentage": "76.50%", "elapsed_time": "28m 38s", "remaining_time": "8m 47s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162958}
+{"loss": 0.36229672, "grad_norm": 0.88381386, "learning_rate": 2.585e-05, "token_acc": 0.89037298, "epoch": 1.55737705, "global_step/max_steps": "285/366", "percentage": "77.87%", "elapsed_time": "28m 58s", "remaining_time": "8m 14s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163892}
+{"loss": 0.32027121, "grad_norm": 1.03780007, "learning_rate": 2.288e-05, "token_acc": 0.90419708, "epoch": 1.58469945, "global_step/max_steps": "290/366", "percentage": "79.23%", "elapsed_time": "29m 16s", "remaining_time": "7m 40s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165096}
+{"loss": 0.30107689, "grad_norm": 0.89192665, "learning_rate": 2.007e-05, "token_acc": 0.90473027, "epoch": 1.61202186, "global_step/max_steps": "295/366", "percentage": "80.60%", "elapsed_time": "29m 34s", "remaining_time": "7m 7s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166257}
+{"loss": 0.37755942, "grad_norm": 1.09636414, "learning_rate": 1.742e-05, "token_acc": 0.88417705, "epoch": 1.63934426, "global_step/max_steps": "300/366", "percentage": "81.97%", "elapsed_time": "29m 53s", "remaining_time": "6m 34s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.16729}
+{"eval_loss": 0.64178872, "eval_runtime": 46.8137, "eval_samples_per_second": 26.467, "eval_steps_per_second": 0.833, "eval_token_acc": 0.82412385, "epoch": 1.63934426, "global_step/max_steps": "300/366", "percentage": "81.97%", "elapsed_time": "30m 40s", "remaining_time": "6m 44s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163034}
+{"loss": 0.37487755, "grad_norm": 0.86775172, "learning_rate": 1.495e-05, "token_acc": 0.88703124, "epoch": 1.66666667, "global_step/max_steps": "305/366", "percentage": "83.33%", "elapsed_time": "31m 0s", "remaining_time": "6m 12s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163931}
+{"loss": 0.34674411, "grad_norm": 0.89181811, "learning_rate": 1.265e-05, "token_acc": 0.89381542, "epoch": 1.69398907, "global_step/max_steps": "310/366", "percentage": "84.70%", "elapsed_time": "31m 18s", "remaining_time": "5m 39s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164985}
+{"loss": 0.32429404, "grad_norm": 0.97413468, "learning_rate": 1.053e-05, "token_acc": 0.90013037, "epoch": 1.72131148, "global_step/max_steps": "315/366", "percentage": "86.07%", "elapsed_time": "31m 37s", "remaining_time": "5m 7s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165976}
+{"loss": 0.31799629, "grad_norm": 0.83369029, "learning_rate": 8.6e-06, "token_acc": 0.89916382, "epoch": 1.74863388, "global_step/max_steps": "320/366", "percentage": "87.43%", "elapsed_time": "31m 56s", "remaining_time": "4m 35s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166981}
+{"eval_loss": 0.63690734, "eval_runtime": 46.8895, "eval_samples_per_second": 26.424, "eval_steps_per_second": 0.832, "eval_token_acc": 0.82480947, "epoch": 1.74863388, "global_step/max_steps": "320/366", "percentage": "87.43%", "elapsed_time": "32m 43s", "remaining_time": "4m 42s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162992}
+{"loss": 0.33394914, "grad_norm": 0.84270769, "learning_rate": 6.85e-06, "token_acc": 0.89953102, "epoch": 1.77595628, "global_step/max_steps": "325/366", "percentage": "88.80%", "elapsed_time": "33m 2s", "remaining_time": "4m 10s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163949}
+{"loss": 0.31105692, "grad_norm": 0.88417625, "learning_rate": 5.29e-06, "token_acc": 0.90583152, "epoch": 1.80327869, "global_step/max_steps": "330/366", "percentage": "90.16%", "elapsed_time": "33m 21s", "remaining_time": "3m 38s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164888}
+{"loss": 0.29880593, "grad_norm": 0.97848785, "learning_rate": 3.94e-06, "token_acc": 0.90583965, "epoch": 1.83060109, "global_step/max_steps": "335/366", "percentage": "91.53%", "elapsed_time": "33m 41s", "remaining_time": "3m 7s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165719}
+{"loss": 0.3232415, "grad_norm": 1.07746804, "learning_rate": 2.77e-06, "token_acc": 0.90018892, "epoch": 1.8579235, "global_step/max_steps": "340/366", "percentage": "92.90%", "elapsed_time": "33m 59s", "remaining_time": "2m 35s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166695}
+{"eval_loss": 0.63737231, "eval_runtime": 46.5271, "eval_samples_per_second": 26.63, "eval_steps_per_second": 0.838, "eval_token_acc": 0.82516718, "epoch": 1.8579235, "global_step/max_steps": "340/366", "percentage": "92.90%", "elapsed_time": "34m 46s", "remaining_time": "2m 39s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.162977}
+{"loss": 0.33315778, "grad_norm": 1.08617711, "learning_rate": 1.81e-06, "token_acc": 0.89642322, "epoch": 1.8852459, "global_step/max_steps": "345/366", "percentage": "94.26%", "elapsed_time": "35m 3s", "remaining_time": "2m 8s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164003}
+{"loss": 0.32831001, "grad_norm": 0.78474462, "learning_rate": 1.05e-06, "token_acc": 0.89639776, "epoch": 1.91256831, "global_step/max_steps": "350/366", "percentage": "95.63%", "elapsed_time": "35m 22s", "remaining_time": "1m 37s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164887}
+{"loss": 0.28941045, "grad_norm": 1.01080036, "learning_rate": 5e-07, "token_acc": 0.91258311, "epoch": 1.93989071, "global_step/max_steps": "355/366", "percentage": "96.99%", "elapsed_time": "35m 38s", "remaining_time": "1m 6s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.165976}
+{"loss": 0.30536501, "grad_norm": 0.98041725, "learning_rate": 1.5e-07, "token_acc": 0.90193578, "epoch": 1.96721311, "global_step/max_steps": "360/366", "percentage": "98.36%", "elapsed_time": "35m 57s", "remaining_time": "35s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.166835}
+{"eval_loss": 0.63765621, "eval_runtime": 46.8617, "eval_samples_per_second": 26.439, "eval_steps_per_second": 0.832, "eval_token_acc": 0.8250231, "epoch": 1.96721311, "global_step/max_steps": "360/366", "percentage": "98.36%", "elapsed_time": "36m 44s", "remaining_time": "36s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.163288}
+{"loss": 0.32302711, "grad_norm": 1.01541507, "learning_rate": 0.0, "token_acc": 0.90018592, "epoch": 1.99453552, "global_step/max_steps": "365/366", "percentage": "99.73%", "elapsed_time": "37m 4s", "remaining_time": "6s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.164074}
+{"eval_loss": 0.63769114, "eval_runtime": 46.8458, "eval_samples_per_second": 26.448, "eval_steps_per_second": 0.833, "eval_token_acc": 0.82501317, "epoch": 2.0, "global_step/max_steps": "366/366", "percentage": "100.00%", "elapsed_time": "37m 53s", "remaining_time": "0s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.160992}
+{"train_runtime": 2275.8777, "train_samples_per_second": 5.129, "train_steps_per_second": 0.161, "total_flos": 2.177813773895598e+17, "train_loss": 0.50223168, "epoch": 2.0, "global_step/max_steps": "366/366", "percentage": "100.00%", "elapsed_time": "37m 54s", "remaining_time": "0s", "memory(GiB)": 54.76, "train_speed(iter/s)": 0.160934}
+{"model_parameter_info": "PeftModelForCausalLM: 8856.7114M Params (89.5877M Trainable [1.0115%]), 0.0001M Buffers.", "last_model_checkpoint": "/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758/checkpoint-366", "best_model_checkpoint": "/root/quantum-assistant/outputs/train/exp-rslora-r32-2/v0-20251209-025758/checkpoint-160", "best_metric": 0.61867183, "global_step": 366, "log_history": [{"loss": 1.2790647745132446, "grad_norm": 7.516477108001709, "learning_rate": 1e-05, "token_acc": 0.7818756585879874, "epoch": 0.00546448087431694, "step": 1}, {"loss": 1.1431974172592163, "grad_norm": 2.4674792289733887, "learning_rate": 5e-05, "token_acc": 0.772373993577512, "epoch": 0.0273224043715847, "step": 5}, {"loss": 0.9257817268371582, "grad_norm": 1.4030452966690063, "learning_rate": 0.0001, "token_acc": 0.7604314994606257, "epoch": 0.0546448087431694, "step": 10}, {"loss": 0.8572605133056641, "grad_norm": 1.33702552318573, "learning_rate": 0.00015000000000000001, "token_acc": 0.766743648960739, "epoch": 0.08196721311475409, "step": 15}, {"loss": 0.6235889434814453, "grad_norm": 1.4701560735702515, "learning_rate": 0.0002, "token_acc": 0.8272145047808046, "epoch": 0.1092896174863388, "step": 20}, {"eval_loss": 0.7344434261322021, "eval_runtime": 47.1079, "eval_samples_per_second": 26.301, "eval_steps_per_second": 0.828, "eval_token_acc": 0.7949850455588787, "epoch": 0.1092896174863388, "step": 20}, {"loss": 0.6710491180419922, "grad_norm": 0.8882436156272888, "learning_rate": 0.00019989696542885188, "token_acc": 0.8127437697581804, "epoch": 0.1366120218579235, "step": 25}, {"loss": 0.6839078426361084, "grad_norm": 1.1426690816879272, "learning_rate": 0.00019958807403786452, "token_acc": 0.8091872791519434, "epoch": 0.16393442622950818, "step": 30}, {"loss": 0.638311767578125, "grad_norm": 0.8387622833251953, "learning_rate": 0.000199073962356878, "token_acc": 0.825588623824695, "epoch": 0.1912568306010929, "step": 35}, {"loss": 0.7615083694458008, "grad_norm": 1.068940281867981, "learning_rate": 0.00019835568981142374, "token_acc": 0.7898940254953156, "epoch": 0.2185792349726776, "step": 40}, {"eval_loss": 0.6962721347808838, "eval_runtime": 46.9273, "eval_samples_per_second": 26.403, "eval_steps_per_second": 0.831, "eval_token_acc": 0.8038185991514224, "epoch": 0.2185792349726776, "step": 40}, {"loss": 0.6525201797485352, "grad_norm": 0.8042981028556824, "learning_rate": 0.0001974347365395755, "token_acc": 0.8161739721372749, "epoch": 0.2459016393442623, "step": 45}, {"loss": 0.6602028846740723, "grad_norm": 1.0532574653625488, "learning_rate": 0.00019631300034184153, "token_acc": 0.8115324923395281, "epoch": 0.273224043715847, "step": 50}, {"loss": 0.6866428852081299, "grad_norm": 0.9879045486450195, "learning_rate": 0.00019499279277038336, "token_acc": 0.8062055954722434, "epoch": 0.3005464480874317, "step": 55}, {"loss": 0.6726794242858887, "grad_norm": 0.8924447298049927, "learning_rate": 0.00019347683436561999, "token_acc": 0.8107636309612242, "epoch": 0.32786885245901637, "step": 60}, {"eval_loss": 0.6777984499931335, "eval_runtime": 47.0363, "eval_samples_per_second": 26.341, "eval_steps_per_second": 0.829, "eval_token_acc": 0.8085781853953239, "epoch": 0.32786885245901637, "step": 60}, {"loss": 0.6699145317077637, "grad_norm": 0.9476580023765564, "learning_rate": 0.0001917682490500337, "token_acc": 0.8170155864777903, "epoch": 0.3551912568306011, "step": 65}, {"loss": 0.6619604110717774, "grad_norm": 0.727776050567627, "learning_rate": 0.00018987055769072972, "token_acc": 0.813844805763801, "epoch": 0.3825136612021858, "step": 70}, {"loss": 0.6429046154022217, "grad_norm": 0.8572030067443848, "learning_rate": 0.00018778767084401556, "token_acc": 0.818294051627385, "epoch": 0.4098360655737705, "step": 75}, {"loss": 0.6941839694976807, "grad_norm": 0.8246414065361023, "learning_rate": 0.0001855238806969513, "token_acc": 0.8030803488587864, "epoch": 0.4371584699453552, "step": 80}, {"eval_loss": 0.6597629189491272, "eval_runtime": 46.8466, "eval_samples_per_second": 26.448, "eval_steps_per_second": 0.833, "eval_token_acc": 0.8126074384681883, "epoch": 0.4371584699453552, "step": 80}, {"loss": 0.7221175193786621, "grad_norm": 0.8452114462852478, "learning_rate": 0.00018308385222247638, "token_acc": 0.7938556379771703, "epoch": 0.4644808743169399, "step": 85}, {"loss": 0.6068331718444824, "grad_norm": 0.8799559473991394, "learning_rate": 0.0001804726135663399, "token_acc": 0.8304502781641169, "epoch": 0.4918032786885246, "step": 90}, {"loss": 0.6669285774230957, "grad_norm": 0.8048533201217651, "learning_rate": 0.00017769554568564387, "token_acc": 0.8149484164201717, "epoch": 0.5191256830601093, "step": 95}, {"loss": 0.6111996173858643, "grad_norm": 0.8785684704780579, "learning_rate": 0.00017475837126035106, "token_acc": 0.8272045408473545, "epoch": 0.546448087431694, "step": 100}, {"eval_loss": 0.6528913974761963, "eval_runtime": 46.5122, "eval_samples_per_second": 26.638, "eval_steps_per_second": 0.838, "eval_token_acc": 0.8137849143970032, "epoch": 0.546448087431694, "step": 100}, {"loss": 0.641059684753418, "grad_norm": 0.8876831531524658, "learning_rate": 0.00017166714290060736, "token_acc": 0.8206215809085864, "epoch": 0.5737704918032787, "step": 105}, {"loss": 0.6174559593200684, "grad_norm": 0.7827315330505371, "learning_rate": 0.00016842823067418018, "token_acc": 0.8254029327519131, "epoch": 0.6010928961748634, "step": 110}, {"loss": 0.6313720703125, "grad_norm": 0.758630633354187, "learning_rate": 0.0001650483089797142, "token_acc": 0.8157943543899996, "epoch": 0.6284153005464481, "step": 115}, {"loss": 0.6850579261779786, "grad_norm": 0.8770219087600708, "learning_rate": 0.0001615343427928555, "token_acc": 0.811273236058365, "epoch": 0.6557377049180327, "step": 120}, {"eval_loss": 0.6429011225700378, "eval_runtime": 46.8791, "eval_samples_per_second": 26.43, "eval_steps_per_second": 0.832, "eval_token_acc": 0.8158268663241884, "epoch": 0.6557377049180327, "step": 120}, {"loss": 0.587221908569336, "grad_norm": 0.7737534642219543, "learning_rate": 0.00015789357331358594, "token_acc": 0.8313492808592137, "epoch": 0.6830601092896175, "step": 125}, {"loss": 0.6085454940795898, "grad_norm": 1.0097602605819702, "learning_rate": 0.0001541335030443444, "token_acc": 0.8255246979012084, "epoch": 0.7103825136612022, "step": 130}, {"loss": 0.5239658355712891, "grad_norm": 0.7576715350151062, "learning_rate": 0.00015026188032968448, "token_acc": 0.8510056426766366, "epoch": 0.7377049180327869, "step": 135}, {"loss": 0.6153683662414551, "grad_norm": 0.8559680581092834, "learning_rate": 0.0001462866833893272, "token_acc": 0.822461361321465, "epoch": 0.7650273224043715, "step": 140}, {"eval_loss": 0.6382603049278259, "eval_runtime": 46.6167, "eval_samples_per_second": 26.578, "eval_steps_per_second": 0.837, "eval_token_acc": 0.8174018024821391, "epoch": 0.7650273224043715, "step": 140}, {"loss": 0.5478519916534423, "grad_norm": 0.8601051568984985, "learning_rate": 0.00014221610387751214, "token_acc": 0.8381034872335618, "epoch": 0.7923497267759563, "step": 145}, {"loss": 0.6614881992340088, "grad_norm": 0.7826522588729858, "learning_rate": 0.00013805853000252585, "token_acc": 0.8124669437098602, "epoch": 0.819672131147541, "step": 150}, {"loss": 0.6105924606323242, "grad_norm": 0.9274362325668335, "learning_rate": 0.00013382252924119272, "token_acc": 0.8269658611430764, "epoch": 0.8469945355191257, "step": 155}, {"loss": 0.632401704788208, "grad_norm": 0.9226640462875366, "learning_rate": 0.0001295168306839494, "token_acc": 0.8244192328471097, "epoch": 0.8743169398907104, "step": 160}, {"eval_loss": 0.6263616681098938, "eval_runtime": 46.5309, "eval_samples_per_second": 26.627, "eval_steps_per_second": 0.838, "eval_token_acc": 0.8203976589592504, "epoch": 0.8743169398907104, "step": 160}, {"loss": 0.5990039348602295, "grad_norm": 0.8398458361625671, "learning_rate": 0.0001251503070468826, "token_acc": 0.8258221211303443, "epoch": 0.9016393442622951, "step": 165}, {"loss": 0.6273468494415283, "grad_norm": 0.8892788887023926, "learning_rate": 0.00012073195638779943, "token_acc": 0.8204203196442804, "epoch": 0.9289617486338798, "step": 170}, {"loss": 0.590123987197876, "grad_norm": 0.7530755996704102, "learning_rate": 0.0001162708835640067, "token_acc": 0.8310831083108311, "epoch": 0.9562841530054644, "step": 175}, {"loss": 0.5976090431213379, "grad_norm": 0.9162634015083313, "learning_rate": 0.00011177628147000961, "token_acc": 0.8295438295438295, "epoch": 0.9836065573770492, "step": 180}, {"eval_loss": 0.6186718344688416, "eval_runtime": 47.2246, "eval_samples_per_second": 26.236, "eval_steps_per_second": 0.826, "eval_token_acc": 0.8218732300599171, "epoch": 0.9836065573770492, "step": 180}, {"loss": 0.5480464458465576, "grad_norm": 0.7956599593162537, "learning_rate": 0.00010725741209379285, "token_acc": 0.8472180707972159, "epoch": 1.010928961748634, "step": 185}, {"loss": 0.41285219192504885, "grad_norm": 0.8504087924957275, "learning_rate": 0.00010272358743072152, "token_acc": 0.8754102158217829, "epoch": 1.0382513661202186, "step": 190}, {"loss": 0.3537081241607666, "grad_norm": 0.9194325804710388, "learning_rate": 9.818415029439201e-05, "token_acc": 0.8963255076843696, "epoch": 1.0655737704918034, "step": 195}, {"loss": 0.33952970504760743, "grad_norm": 0.9619770646095276, "learning_rate": 9.364845506397624e-05, "token_acc": 0.8953101361573373, "epoch": 1.092896174863388, "step": 200}, {"eval_loss": 0.6519283652305603, "eval_runtime": 46.9396, "eval_samples_per_second": 26.396, "eval_steps_per_second": 0.831, "eval_token_acc": 0.8215900396466579, "epoch": 1.092896174863388, "step": 200}, {"loss": 0.35062572956085203, "grad_norm": 0.8512245416641235, "learning_rate": 8.91258484077327e-05, "token_acc": 0.8883580849517084, "epoch": 1.1202185792349726, "step": 205}, {"loss": 0.33119356632232666, "grad_norm": 0.8422494530677795, "learning_rate": 8.462565002240732e-05, "token_acc": 0.897713258870371, "epoch": 1.1475409836065573, "step": 210}, {"loss": 0.3076974153518677, "grad_norm": 1.0778322219848633, "learning_rate": 8.01571334282144e-05, "token_acc": 0.9026581649552832, "epoch": 1.174863387978142, "step": 215}, {"loss": 0.34732604026794434, "grad_norm": 1.004669189453125, "learning_rate": 7.572950685897294e-05, "token_acc": 0.8883448456763137, "epoch": 1.2021857923497268, "step": 220}, {"eval_loss": 0.6518204808235168, "eval_runtime": 46.6705, "eval_samples_per_second": 26.548, "eval_steps_per_second": 0.836, "eval_token_acc": 0.8216049444052504, "epoch": 1.2021857923497268, "step": 220}, {"loss": 0.29864633083343506, "grad_norm": 0.7113569378852844, "learning_rate": 7.135189428677829e-05, "token_acc": 0.9103727538956579, "epoch": 1.2295081967213115, "step": 225}, {"loss": 0.33050739765167236, "grad_norm": 0.9444766044616699, "learning_rate": 6.703331662031098e-05, "token_acc": 0.8991153251081049, "epoch": 1.2568306010928962, "step": 230}, {"loss": 0.36874344348907473, "grad_norm": 0.9812134504318237, "learning_rate": 6.278267311552772e-05, "token_acc": 0.8841693397722866, "epoch": 1.2841530054644807, "step": 235}, {"loss": 0.3378627300262451, "grad_norm": 1.1992689371109009, "learning_rate": 5.860872303704089e-05, "token_acc": 0.8950307091010609, "epoch": 1.3114754098360657, "step": 240}, {"eval_loss": 0.652209997177124, "eval_runtime": 46.8505, "eval_samples_per_second": 26.446, "eval_steps_per_second": 0.832, "eval_token_acc": 0.8219427856000159, "epoch": 1.3114754098360657, "step": 240}, {"loss": 0.3598504543304443, "grad_norm": 1.10702383518219, "learning_rate": 5.452006760797708e-05, "token_acc": 0.8884325804243669, "epoch": 1.3387978142076502, "step": 245}, {"loss": 0.33976819515228274, "grad_norm": 0.9865630269050598, "learning_rate": 5.052513228551048e-05, "token_acc": 0.8951826363155109, "epoch": 1.366120218579235, "step": 250}, {"loss": 0.34928138256073, "grad_norm": 0.9960690140724182, "learning_rate": 4.663214939859529e-05, "token_acc": 0.8938934603724557, "epoch": 1.3934426229508197, "step": 255}, {"loss": 0.32947587966918945, "grad_norm": 1.0176554918289185, "learning_rate": 4.2849141183676365e-05, "token_acc": 0.8972827214270899, "epoch": 1.4207650273224044, "step": 260}, {"eval_loss": 0.6426578164100647, "eval_runtime": 46.6918, "eval_samples_per_second": 26.536, "eval_steps_per_second": 0.835, "eval_token_acc": 0.8234183567006826, "epoch": 1.4207650273224044, "step": 260}, {"loss": 0.3543287754058838, "grad_norm": 1.0032607316970825, "learning_rate": 3.9183903253335185e-05, "token_acc": 0.889045292705142, "epoch": 1.4480874316939891, "step": 265}, {"loss": 0.3612166404724121, "grad_norm": 0.8782271146774292, "learning_rate": 3.5643988531937924e-05, "token_acc": 0.8886253016483292, "epoch": 1.4754098360655736, "step": 270}, {"loss": 0.2726799488067627, "grad_norm": 0.7909438014030457, "learning_rate": 3.2236691691388974e-05, "token_acc": 0.9156046073665309, "epoch": 1.5027322404371586, "step": 275}, {"loss": 0.34633607864379884, "grad_norm": 0.8114061951637268, "learning_rate": 2.8969034119063176e-05, "token_acc": 0.8960518085145528, "epoch": 1.530054644808743, "step": 280}, {"eval_loss": 0.6389462351799011, "eval_runtime": 46.8001, "eval_samples_per_second": 26.474, "eval_steps_per_second": 0.833, "eval_token_acc": 0.8242232136646827, "epoch": 1.530054644808743, "step": 280}, {"loss": 0.36229672431945803, "grad_norm": 0.8838138580322266, "learning_rate": 2.5847749448892944e-05, "token_acc": 0.8903729838709677, "epoch": 1.5573770491803278, "step": 285}, {"loss": 0.3202712059020996, "grad_norm": 1.0378000736236572, "learning_rate": 2.287926968542674e-05, "token_acc": 0.9041970802919708, "epoch": 1.5846994535519126, "step": 290}, {"loss": 0.30107688903808594, "grad_norm": 0.891926646232605, "learning_rate": 2.0069711949452385e-05, "token_acc": 0.9047302684028701, "epoch": 1.6120218579234973, "step": 295}, {"loss": 0.3775594234466553, "grad_norm": 1.096364140510559, "learning_rate": 1.742486587249873e-05, "token_acc": 0.8841770474423536, "epoch": 1.639344262295082, "step": 300}, {"eval_loss": 0.6417887210845947, "eval_runtime": 46.8137, "eval_samples_per_second": 26.467, "eval_steps_per_second": 0.833, "eval_token_acc": 0.8241238486073987, "epoch": 1.639344262295082, "step": 300}, {"loss": 0.3748775482177734, "grad_norm": 0.8677517175674438, "learning_rate": 1.4950181666191599e-05, "token_acc": 0.8870312443821234, "epoch": 1.6666666666666665, "step": 305}, {"loss": 0.34674410820007323, "grad_norm": 0.891818106174469, "learning_rate": 1.2650758891049463e-05, "token_acc": 0.8938154209052042, "epoch": 1.6939890710382515, "step": 310}, {"loss": 0.3242940425872803, "grad_norm": 0.9741346836090088, "learning_rate": 1.053133594786283e-05, "token_acc": 0.9001303677952041, "epoch": 1.721311475409836, "step": 315}, {"loss": 0.31799628734588625, "grad_norm": 0.8336902856826782, "learning_rate": 8.596280313312355e-06, "token_acc": 0.8991638160395287, "epoch": 1.748633879781421, "step": 320}, {"eval_loss": 0.6369073390960693, "eval_runtime": 46.8895, "eval_samples_per_second": 26.424, "eval_steps_per_second": 0.832, "eval_token_acc": 0.824809467502658, "epoch": 1.748633879781421, "step": 320}, {"loss": 0.3339491367340088, "grad_norm": 0.8427076935768127, "learning_rate": 6.849579539947115e-06, "token_acc": 0.8995310176535631, "epoch": 1.7759562841530054, "step": 325}, {"loss": 0.311056923866272, "grad_norm": 0.8841762542724609, "learning_rate": 5.294833039069269e-06, "token_acc": 0.9058315166614762, "epoch": 1.8032786885245902, "step": 330}, {"loss": 0.2988059282302856, "grad_norm": 0.9784878492355347, "learning_rate": 3.935244663458048e-06, "token_acc": 0.9058396498345255, "epoch": 1.830601092896175, "step": 335}, {"loss": 0.3232414960861206, "grad_norm": 1.0774680376052856, "learning_rate": 2.7736161052178356e-06, "token_acc": 0.9001889168765743, "epoch": 1.8579234972677594, "step": 340}, {"eval_loss": 0.6373723149299622, "eval_runtime": 46.5271, "eval_samples_per_second": 26.63, "eval_steps_per_second": 0.838, "eval_token_acc": 0.8251671817088803, "epoch": 1.8579234972677594, "step": 340}, {"loss": 0.3331577777862549, "grad_norm": 1.086177110671997, "learning_rate": 1.8123411223552544e-06, "token_acc": 0.8964232169629471, "epoch": 1.8852459016393444, "step": 345}, {"loss": 0.3283100128173828, "grad_norm": 0.7847446203231812, "learning_rate": 1.053400605982613e-06, "token_acc": 0.8963977599515666, "epoch": 1.9125683060109289, "step": 350}, {"loss": 0.2894104480743408, "grad_norm": 1.0108003616333008, "learning_rate": 4.983584983125345e-07, "token_acc": 0.9125831076089633, "epoch": 1.9398907103825138, "step": 355}, {"loss": 0.3053650140762329, "grad_norm": 0.9804172515869141, "learning_rate": 1.4835856985568887e-07, "token_acc": 0.901935783006839, "epoch": 1.9672131147540983, "step": 360}, {"eval_loss": 0.6376562118530273, "eval_runtime": 46.8617, "eval_samples_per_second": 26.439, "eval_steps_per_second": 0.832, "eval_token_acc": 0.8250231023758186, "epoch": 1.9672131147540983, "step": 360}, {"loss": 0.32302711009979246, "grad_norm": 1.015415072441101, "learning_rate": 4.1220624626636675e-09, "token_acc": 0.9001859210382414, "epoch": 1.994535519125683, "step": 365}, {"eval_loss": 0.6376911401748657, "eval_runtime": 46.8458, "eval_samples_per_second": 26.448, "eval_steps_per_second": 0.833, "eval_token_acc": 0.8250131658700901, "epoch": 2.0, "step": 366}, {"train_runtime": 2275.8777, "train_samples_per_second": 5.129, "train_steps_per_second": 0.161, "total_flos": 2.177813773895598e+17, "train_loss": 0.5022316773732504, "epoch": 2.0, "step": 366}], "memory": 54.763671875}

training_artifacts/v0-20251209-025758/runs/events.out.tfevents.1765249089.837c4d588d5e.556267.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afe518b92171710201cb1b38eb7f3250c87951d6fb308765fa4efbbf65cc575c
+size 33383