================================================================================   
BARTLEBY FULL FINETUNE _ 16-BIT _ AUTO TEMPLATE+MASK DETECT _ LAST-ANSWER MULTITURN
================================================================================   
MODEL      : unsloth/Qwen3-1.7B                                                    
DATA       : data/training_data_v2_filtered.jsonl                                  
GOLD       : data/gold_seed_training_data_sosts.jsonl                              
OUTPUT     : staeiou/bartleby-qwen3-1.7b_v5                                        
CACHE_DIR  : /workspace/.cache/huggingface/datasets                                
SEQ        : 1024                                                                  
PACKING    : False                                                                 
LOAD_4BIT  : False (forced 16-bit base)                                            
FULL_FT    : True                                                                  
REMOTE_CODE: False                                                                 
FAMILY     : qwen                                                                  
TRL_COMPAT : ConstantLengthDataset patched=True                                    
ADAPTERS   : disabled                                                              
TRAIN      : bs=2 grad_accum=16 eff_bs=32                                          
EPOCHS     : 2.0                                                                   
LR         : 1e-05 scheduler=cosine warmup=0.03 weight_decay=0.05 max_grad_norm=1.0
MULTITURN  : num=0 max_turns=5 (only last assistant supervised)                    
GOLD_REPEAT: 5                                                                     
GPU        : Single GPU (CUDA_VISIBLE_DEVICES=0)                                   
================================================================================   
                                                                                   
[1/7] Loading base model...                                                        
==((====))==  Unsloth 2026.3.5: Fast Qwen3 patching. Transformers: 5.3.0. vLLM: 0.1
3.0.                                                                               
   \\   /|    NVIDIA RTX 5000 Ada Generation. Num GPUs = 1. Max memory: 31.475 GB. 
Platform: Linux.                                                                   
O^O/ \_/ \    Torch: 2.9.0+cu128. CUDA: 8.9. CUDA Toolkit: 12.8. Triton: 3.5.0     
\        /    Bfloat16 = TRUE. FA [Xformers = 0.0.33.post1. FA2 = False]            "-____-"     Free license: http://github.com/unslothai/unsloth