File size: 3,708 Bytes
1e7d587 d1c8896 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | 2025-12-11 11:12:17,843 __main__ <module> [INFO] OUT_DIR set to outputs/2025-12-11/11-12-17
2025-12-11 11:12:17,844 __main__ <module> [INFO] Args: app/src/S3_8_dpo_human.py --dpo.per_device_train_batch_size 32 --dpo.per_device_eval_batch_size 32 --dpo.gradient_accumulation_steps 32 --dpo.push_to_hub --model.model_name_or_path tokyotech-llm/Llama-3.1-Swallow-8B-v0.5
2025-12-11 11:12:18,587 accelerate.utils.modeling get_balanced_memory [INFO] We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).
2025-12-11 11:12:31,964 __main__ load_model [INFO] Model loaded from tokyotech-llm/Llama-3.1-Swallow-8B-v0.5
LlamaForCausalLM(
(model): LlamaModel(
(embed_tokens): Embedding(128256, 4096)
(layers): ModuleList(
(0-31): 32 x LlamaDecoderLayer(
(self_attn): LlamaAttention(
(q_proj): Linear4bit(in_features=4096, out_features=4096, bias=False)
(k_proj): Linear4bit(in_features=4096, out_features=1024, bias=False)
(v_proj): Linear4bit(in_features=4096, out_features=1024, bias=False)
(o_proj): Linear4bit(in_features=4096, out_features=4096, bias=False)
)
(mlp): LlamaMLP(
(gate_proj): Linear4bit(in_features=4096, out_features=14336, bias=False)
(up_proj): Linear4bit(in_features=4096, out_features=14336, bias=False)
(down_proj): Linear4bit(in_features=14336, out_features=4096, bias=False)
(act_fn): SiLUActivation()
)
(input_layernorm): LlamaRMSNorm((4096,), eps=1e-05)
(post_attention_layernorm): LlamaRMSNorm((4096,), eps=1e-05)
)
)
(norm): LlamaRMSNorm((4096,), eps=1e-05)
(rotary_emb): LlamaRotaryEmbedding()
)
(lm_head): Linear(in_features=4096, out_features=128256, bias=False)
)
2025-12-11 11:12:34,958 __main__ load_jwtd [INFO] Dataset loaded, N=2554
2025-12-11 11:12:36,075 __main__ load_jwtd [INFO] Dataset loaded, N=542
2025-12-11 11:12:36,075 __main__ main [INFO] Filtered dataset: train 2554 rows, eval 542 rows
2025-12-11 11:12:47,525 __main__ add_train_str [INFO] pre_str:post_str = 469:2085 = 0.18:0.82
2025-12-11 11:12:59,077 __main__ add_train_str [INFO] pre_str:post_str = 110:432 = 0.20:0.80
2025-12-11 11:13:02,272 __main__ main [INFO] wandb initialized
2025-12-11 11:13:07,268 __main__ main [INFO] Starting DPO training with DPOTrainer
2025-12-11 11:13:32,682 root evaluate_probability_ratio [INFO] Results epoch 0: outputs/2025-12-11/11-12-17/probability_ratio_epoch_0.json
2025-12-11 11:15:24,572 root evaluate_probability_ratio [INFO] Results epoch 1: outputs/2025-12-11/11-12-17/probability_ratio_epoch_1.json
2025-12-11 11:17:15,106 root evaluate_probability_ratio [INFO] Results epoch 2: outputs/2025-12-11/11-12-17/probability_ratio_epoch_2.json
2025-12-11 11:19:05,077 root evaluate_probability_ratio [INFO] Results epoch 3: outputs/2025-12-11/11-12-17/probability_ratio_epoch_3.json
2025-12-11 11:20:57,951 root evaluate_probability_ratio [INFO] Results epoch 4: outputs/2025-12-11/11-12-17/probability_ratio_epoch_4.json
2025-12-11 11:22:49,543 root evaluate_probability_ratio [INFO] Results epoch 5: outputs/2025-12-11/11-12-17/probability_ratio_epoch_5.json
2025-12-11 11:24:41,323 root evaluate_probability_ratio [INFO] Results epoch 6: outputs/2025-12-11/11-12-17/probability_ratio_epoch_6.json
2025-12-11 11:26:32,747 root evaluate_probability_ratio [INFO] Results epoch 7: outputs/2025-12-11/11-12-17/probability_ratio_epoch_7.json
2025-12-11 11:28:25,903 root evaluate_probability_ratio [INFO] Results epoch 8: outputs/2025-12-11/11-12-17/probability_ratio_epoch_8.json
|