base_model: Qwen/Qwen2.5-Coder-7B-Instruct
model_type: Qwen2ForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false

# Pre-tokenized datasets produced by scripts/preprocess_diff_mask_chat.py
# (from felixwangg/prime_vul_plus_splitted which has a fixed validation split).
# Columns: input_ids, attention_mask, labels, diff_mask.
# Labels are already -100 for non-assistant tokens; axolotl keeps them as-is.
datasets:
  - path: felixwangg/prime_vul_plus_splitted_token_diff_mask_skip_indent_ctx0_chat_v2
    type: pretokenized
    split: train
test_datasets:
  - path: felixwangg/prime_vul_plus_splitted_token_diff_mask_skip_indent_ctx0_chat_v2
    type: pretokenized
    split: validation
dataset_prepared_path: /home/tkwang/scratch/SecSteer-v2/axolotl-datasets/lora/Qwen2.5-Coder-7B/prime_vul_plus_splitted_token_diff_mask_skip_indent_ctx0_chat_v2
val_set_size: 0
output_dir: /home/tkwang/scratch/SecSteer-v2/axolotl-outputs/lora/Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2
sequence_len: 4096
sample_packing: false
eval_sample_packing: false
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 16
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
merge_lora: true

wandb_project: diff-mask-sft-primevul-ctx-0
wandb_entity: wtkuan
wandb_watch: "false"
wandb_name: Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2
wandb_log_model: "false"

gradient_accumulation_steps: 8
micro_batch_size: 4
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 4e-05

bf16: true
tf32: false

gradient_checkpointing: true
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

num_epochs: 1
warmup_ratio: 0.1
early_stopping_patience: 1000
eval_steps: 15
save_steps: 15
save_total_limit: 1000
load_best_model_at_end: true

weight_decay: 0.02
special_tokens:

# Diff-mask weighted loss: CE(logit_t, label_t) * (1 + alpha * diff_mask_{t+1})
# Security-sensitive tokens (diff_mask=1) get weight (1 + diff_mask_alpha).
# Requires PYTHONPATH to include the repo root so diff_mask_trainer is importable.
diff_mask_alpha: 1.0

plugins:
  - diff_mask_trainer.plugin.DiffMaskPlugin
  # - sec_bench_callback.SecBenchPlugin

home/tkwang/scratch/SecSteer-v2/axolotl-outputs/lora/Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2

This model is a fine-tuned version of Qwen/Qwen2.5-Coder-7B-Instruct on the felixwangg/prime_vul_plus_splitted_token_diff_mask_skip_indent_ctx0_chat_v2 dataset. It achieves the following results on the evaluation set:

Loss: 0.7659
Ppl: 2.1508
Memory/max Active (gib): 42.7
Memory/max Allocated (gib): 42.7
Memory/device Reserved (gib): 61.19

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 4e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 8
total_train_batch_size: 64
total_eval_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 5
training_steps: 57

Training results

Training Loss	Epoch	Step	Validation Loss	Ppl	Active (gib)	Allocated (gib)	Reserved (gib)
No log	0	0	0.8682	2.3826	42.36	42.36	51.15
7.1944	0.2632	15	0.8215	2.274	42.7	42.7	60.61
6.5999	0.5263	30	0.7745	2.1696	42.7	42.7	61.19
6.1178	0.7895	45	0.7659	2.1508	42.7	42.7	61.19

Framework versions

PEFT 0.18.1
Transformers 4.57.6
Pytorch 2.10.0+cu128
Datasets 4.5.0
Tokenizers 0.22.2

felixwangg/Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2

home/tkwang/scratch/SecSteer-v2/axolotl-outputs/lora/Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

felixwangg/
Qwen2.5-Coder-7B-sft-plus-alpha-1-token-diff-ctx0-v2