output-fizz
This model is a fine-tuned version of Qwen/Qwen3.5-9B.
W&B run: https://wandb.ai/cooawoo-personal/huggingface/runs/r719jwwe
Training procedure
Hyperparameters
| Parameter |
Value |
| Learning rate |
0.0002 |
| LR scheduler |
SchedulerType.COSINE |
| Per-device batch size |
2 |
| Gradient accumulation |
8 |
| Effective batch size |
16 |
| Epochs |
2 |
| Max sequence length |
4096 |
| Optimizer |
OptimizerNames.PAGED_ADEMAMIX_8BIT |
| Weight decay |
0.01 |
| Warmup ratio |
0.05 |
| Max gradient norm |
1.0 |
| Precision |
bf16 |
| Loss type |
nll |
| Chunked cross-entropy |
yes |
LoRA configuration
| Parameter |
Value |
| Rank (r) |
128 |
| Alpha |
8 |
| Dropout |
0.05 |
| Target modules |
attn.proj, down_proj, gate_proj, in_proj_a, in_proj_b, in_proj_qkv, in_proj_z, k_proj, linear_fc1, linear_fc2, o_proj, out_proj, q_proj, qkv, up_proj, v_proj |
| rsLoRA |
yes |
| Quantization |
4-bit (nf4) |
Dataset statistics
| Dataset |
Samples |
Total tokens |
Trainable tokens |
| allura-forge/doubao-seed2.0-claude-distill-v1-qwen3.5-format |
3,644 |
7,187,856 |
6,625,087 |
Training config
model_name_or_path: Qwen/Qwen3.5-9B
bf16: true
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
use_liger: true
use_cce: true
max_length: 4096
learning_rate: 0.0002
warmup_ratio: 0.05
weight_decay: 0.01
lr_scheduler_type: cosine
per_device_train_batch_size: 2
gradient_accumulation_steps: 8
optim: paged_ademamix_8bit
max_grad_norm: 1.0
use_peft: true
load_in_4bit: true
lora_r: 128
lora_alpha: 8
lora_dropout: 0.05
use_rslora: true
logging_steps: 1
disable_tqdm: true
save_strategy: steps
save_steps: 500
save_total_limit: 3
report_to: wandb
output_dir: output-fizz
data_config: data.yaml
prepared_dataset: prepared
num_train_epochs: 2
saves_per_epoch: 2
run_name: qwen35-9b-qlora-fizz
Data config
datasets:
- path: allura-forge/doubao-seed2.0-claude-distill-v1-qwen3.5-format
type: conversational
truncation_strategy: drop
shuffle_datasets: true
shuffle_combined: true
shuffle_seed: 42
eval_split: 0.0
split_seed: 42
assistant_only_loss: true
Framework versions
- PEFT 0.18.1
- Loft: 0.1.0
- Transformers: 5.2.0
- Pytorch: 2.10.0
- Datasets: 4.5.0
- Tokenizers: 0.22.2
