Training in progress, step 10

adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "up_proj",
     "v_proj",
-    "gate_proj",
-    "o_proj",
     "down_proj",
+    "o_proj",
+    "q_proj",
     "k_proj",
-    "q_proj"
+    "up_proj",
+    "gate_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:488c788a235fa1bf8dabb096c97892385afe992125107e79145f7e015c39ff9d
+oid sha256:56918846517cc0d46e50c5d931dc4f6a369bc484511c4dd39a37d0da451f8de5
 size 1656903768

args.json

@@ -55,12 +55,14 @@
   "stream": false,
   "stop_words": [],
   "logprobs": false,
-  "ckpt_dir": "/root/highspeedstorage/ag-model-sg/agentic_v5",
+  "ckpt_dir": "/root/dataDisk/output/v0-20250124-130213/checkpoint-5",
   "load_dataset_config": null,
   "lora_modules": [],
   "tuner_backend": "peft",
   "train_type": "lora",
-  "adapters": [],
+  "adapters": [
+    "/root/dataDisk/output/v0-20250124-130213/checkpoint-5"
+  ],
   "seed": 42,
   "model_kwargs": {},
   "load_args": true,
@@ -70,7 +72,7 @@
   "custom_register_path": [],
   "ignore_args_error": false,
   "use_swift_lora": false,
-  "output_dir": "/root/dataDisk/output/v0-20250124-130213",
+  "output_dir": "/root/dataDisk/output/v1-20250125-095524",
   "overwrite_output_dir": false,
   "do_train": false,
   "do_eval": false,
@@ -100,7 +102,7 @@
   "log_level": "passive",
   "log_level_replica": "warning",
   "log_on_each_node": true,
-  "logging_dir": "/root/dataDisk/output/v0-20250124-130213/runs",
+  "logging_dir": "/root/dataDisk/output/v1-20250125-095524/runs",
   "logging_strategy": "steps",
   "logging_first_step": true,
   "logging_steps": 1,
@@ -205,7 +207,7 @@
   "skip_memory_metrics": true,
   "use_legacy_prediction_loop": false,
   "push_to_hub": true,
-  "resume_from_checkpoint": null,
+  "resume_from_checkpoint": "/root/dataDisk/output/v0-20250124-130213/checkpoint-5",
   "hub_model_id": "TheAgenticAI/Agentic-V7",
   "hub_strategy": "every_save",
   "hub_private_repo": true,
@@ -335,8 +337,8 @@
   "local_world_size": 8,
   "model_suffix": "agentic_v5",
   "model_info": "ModelInfo(model_type='llama3_2', model_dir='/root/highspeedstorage/ag-model-sg/agentic_v5', torch_dtype=torch.bfloat16, max_model_len=131072, quant_method=None, quant_bits=None, config=None, task_type=None)",
-  "model_meta": "ModelMeta(model_type='llama3_2', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-1B', hf_model_id='meta-llama/Llama-3.2-1B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B', hf_model_id='meta-llama/Llama-3.2-3B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-1B-Instruct', hf_model_id='meta-llama/Llama-3.2-1B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B-Instruct', hf_model_id='meta-llama/Llama-3.2-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.3-70B-Instruct', hf_model_id='meta-llama/Llama-3.3-70B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', hf_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2', get_function=<function get_model_tokenizer_with_flash_attn at 0x7fb0f7f887c0>, model_arch='llama', architectures=['LlamaForCausalLM'], is_multimodal=False, additional_saved_files=[], torch_dtype=None, ignore_patterns=[], requires=['transformers>=4.45'], tags=[])",
+  "model_meta": "ModelMeta(model_type='llama3_2', model_groups=[ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.2-1B', hf_model_id='meta-llama/Llama-3.2-1B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B', hf_model_id='meta-llama/Llama-3.2-3B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-1B-Instruct', hf_model_id='meta-llama/Llama-3.2-1B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='LLM-Research/Llama-3.2-3B-Instruct', hf_model_id='meta-llama/Llama-3.2-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='LLM-Research/Llama-3.3-70B-Instruct', hf_model_id='meta-llama/Llama-3.3-70B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', hf_model_id='unsloth/Llama-3.3-70B-Instruct-bnb-4bit', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='llama3_2', get_function=<function get_model_tokenizer_with_flash_attn at 0x7fa9a7bc56c0>, model_arch='llama', architectures=['LlamaForCausalLM'], is_multimodal=False, additional_saved_files=[], torch_dtype=None, ignore_patterns=[], requires=['transformers>=4.45'], tags=[])",
   "model_dir": "/root/highspeedstorage/ag-model-sg/agentic_v5",
   "hub": "<class 'swift.hub.hub.HFHub'>",
-  "training_args": "Seq2SeqTrainingArguments(output_dir='/root/dataDisk/output/v0-20250124-130213', 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=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=32, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0001, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/dataDisk/output/v0-20250124-130213/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=5, save_total_limit=2, 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, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=5, dataloader_num_workers=0, dataloader_prefetch_factor=None, past_index=-1, run_name='/root/dataDisk/output/v0-20250124-130213', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'cpu', 'pin_memory': True}, 'overlap_comm': True, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['comet_ml'], 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=True, resume_from_checkpoint=None, hub_model_id='TheAgenticAI/Agentic-V7', hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=True, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': True}, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy='steps', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, acc_strategy='token', sequence_parallel_size=1, check_model=True, train_sampler_random=True, is_encoder_decoder=False, metric_warmup_step=0, train_dataset_sample=-1, fsdp_num=1, acc_steps=1, train_type='lora')"
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/root/dataDisk/output/v1-20250125-095524', 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=1, per_device_eval_batch_size=1, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=32, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=0.0001, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.999, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=3.0, max_steps=-1, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_ratio=0.05, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/dataDisk/output/v1-20250125-095524/runs', logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=5, save_total_limit=2, 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, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=False, eval_steps=5, dataloader_num_workers=0, dataloader_prefetch_factor=None, past_index=-1, run_name='/root/dataDisk/output/v1-20250125-095524', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'cpu', 'pin_memory': True}, 'overlap_comm': True, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=<OptimizerNames.ADAMW_TORCH: 'adamw_torch'>, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['comet_ml'], 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=True, resume_from_checkpoint='/root/dataDisk/output/v0-20250124-130213/checkpoint-5', hub_model_id='TheAgenticAI/Agentic-V7', hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_token=None, hub_private_repo=True, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs={'use_reentrant': True}, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', evaluation_strategy='steps', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=1800, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, dispatch_batches=None, split_batches=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, acc_strategy='token', sequence_parallel_size=1, check_model=True, train_sampler_random=True, is_encoder_decoder=False, metric_warmup_step=0, train_dataset_sample=-1, fsdp_num=1, acc_steps=1, train_type='lora')"
 }

logging.jsonl

@@ -1,6 +1,6 @@
-{"loss": 30.98055649, "token_acc": 0.74572686, "grad_norm": 3.5394487, "learning_rate": 5e-05, "memory(GiB)": 91.49, "train_speed(iter/s)": 0.001155, "epoch": 0.1221374, "global_step/max_steps": "1/24", "percentage": "4.17%", "elapsed_time": "14m 13s", "remaining_time": "5h 27m 5s"}
-{"loss": 32.14899445, "token_acc": 0.72766594, "grad_norm": 1.59590147, "learning_rate": 0.0001, "memory(GiB)": 91.49, "train_speed(iter/s)": 0.001154, "epoch": 0.24427481, "global_step/max_steps": "2/24", "percentage": "8.33%", "elapsed_time": "28m 40s", "remaining_time": "5h 15m 20s"}
-{"loss": 30.30093575, "token_acc": 0.73977445, "grad_norm": 1.98488802, "learning_rate": 9.949e-05, "memory(GiB)": 91.49, "train_speed(iter/s)": 0.001148, "epoch": 0.36641221, "global_step/max_steps": "3/24", "percentage": "12.50%", "elapsed_time": "43m 20s", "remaining_time": "5h 3m 22s"}
-{"loss": 28.31693077, "token_acc": 0.76803195, "grad_norm": 1.28440938, "learning_rate": 9.797e-05, "memory(GiB)": 91.49, "train_speed(iter/s)": 0.001154, "epoch": 0.48854962, "global_step/max_steps": "4/24", "percentage": "16.67%", "elapsed_time": "57m 32s", "remaining_time": "4h 47m 41s"}
-{"loss": 26.89690018, "token_acc": 0.77267505, "grad_norm": 0.29667912, "learning_rate": 9.548e-05, "memory(GiB)": 91.49, "train_speed(iter/s)": 0.001159, "epoch": 0.61068702, "global_step/max_steps": "5/24", "percentage": "20.83%", "elapsed_time": "1h 11m 39s", "remaining_time": "4h 32m 19s"}
-{"eval_loss": 0.78403014, "eval_token_acc": 0.7908435, "eval_runtime": 21.3239, "eval_samples_per_second": 0.985, "eval_steps_per_second": 0.141, "epoch": 0.61068702, "global_step/max_steps": "5/24", "percentage": "20.83%", "elapsed_time": "1h 12m 1s", "remaining_time": "4h 33m 40s"}
+{"loss": 25.51766205, "token_acc": 0.78414556, "grad_norm": 0.18047128, "learning_rate": 9.206e-05, "memory(GiB)": 96.08, "train_speed(iter/s)": 0.006513, "epoch": 0.73282443, "global_step/max_steps": "6/24", "percentage": "25.00%", "elapsed_time": "15m 8s", "remaining_time": "45m 24s"}
+{"loss": 24.12745285, "token_acc": 0.80612886, "grad_norm": 0.11673307, "learning_rate": 8.779e-05, "memory(GiB)": 96.08, "train_speed(iter/s)": 0.003954, "epoch": 0.85496183, "global_step/max_steps": "7/24", "percentage": "29.17%", "elapsed_time": "29m 17s", "remaining_time": "1h 11m 8s"}
+{"loss": 23.98846054, "token_acc": 0.79647431, "grad_norm": 0.11979585, "learning_rate": 8.274e-05, "memory(GiB)": 96.08, "train_speed(iter/s)": 0.003025, "epoch": 0.97709924, "global_step/max_steps": "8/24", "percentage": "33.33%", "elapsed_time": "43m 51s", "remaining_time": "1h 27m 43s"}
+{"loss": 26.9921093, "token_acc": 0.78990856, "grad_norm": 0.10876788, "learning_rate": 7.703e-05, "memory(GiB)": 96.08, "train_speed(iter/s)": 0.00246, "epoch": 1.1221374, "global_step/max_steps": "9/24", "percentage": "37.50%", "elapsed_time": "1h 0m 45s", "remaining_time": "1h 41m 15s"}
+{"loss": 22.41528702, "token_acc": 0.78308229, "grad_norm": 0.1409045, "learning_rate": 7.077e-05, "memory(GiB)": 96.08, "train_speed(iter/s)": 0.002205, "epoch": 1.24427481, "global_step/max_steps": "10/24", "percentage": "41.67%", "elapsed_time": "1h 15m 21s", "remaining_time": "1h 45m 29s"}
+{"eval_loss": 0.6736415, "eval_token_acc": 0.81044838, "eval_runtime": 21.32, "eval_samples_per_second": 0.985, "eval_steps_per_second": 0.141, "epoch": 1.24427481, "global_step/max_steps": "10/24", "percentage": "41.67%", "elapsed_time": "1h 15m 42s", "remaining_time": "1h 45m 59s"}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:89db0b38d1181a25f71011205cc2c85a9f48c2a38877ba554d21d5df908736f9
-size 8120
+oid sha256:6fca9bf0288733d44e6f9680807eb7a3ad176d54ddb4e15eac1e6d5e409bdeeb
+size 8184