base_model: Qwen/Qwen3-8B
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name

quantization_config:
  load_in_4bit: true
  bnb_4bit_quant_type: "nf4"
  bnb_4bit_compute_dtype: "bfloat16"
  bnb_4bit_use_double_quant: true
strict: false

datasets:
  - path: /workspace/outputs/training_data/
    ds_type: json
    data_files: 
        - knowledge_v2.json
    type: chat_template
dataset_prepared_path:
val_set_size: 0.05
output_dir: /workspace/outputs/FT_v2

sequence_len: 1024
sample_packing: true
eval_sample_packing: true


adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 64
lora_dropout: 0.05
lora_target_linear: true

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 0.0002

bf16: auto
tf32: true

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

warmup_ratio: 0.1
evals_per_epoch: 4
saves_per_epoch: 1
weight_decay: 0.0
#fsdp:
#  - full_shard
#  - auto_wrap
#fsdp_config:
#  fsdp_limit_all_gathers: true
#  fsdp_sync_module_states: true
#  fsdp_offload_params: true
#  fsdp_use_orig_params: false
#  fsdp_cpu_ram_efficient_loading: true
#  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
#  fsdp_transformer_layer_cls_to_wrap: Qwen3DecoderLayer
#  fsdp_state_dict_type: FULL_STATE_DICT
#  fsdp_sharding_strategy: FULL_SHARD
special_tokens:

# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

workspace/outputs/FT_v2

This model is a fine-tuned version of Qwen/Qwen3-8B on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1480
Memory/max Active (gib): 17.11
Memory/max Allocated (gib): 17.11
Memory/device Reserved (gib): 20.54

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: 0.0002
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 4
total_train_batch_size: 32
total_eval_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
training_steps: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Active (gib)	Allocated (gib)	Reserved (gib)
No log	0	0	18.8148	17.02	17.02	17.05
1.174	0.2424	2	0.1786	17.11	17.11	20.54
0.4046	0.4848	4	0.1719	17.11	17.11	20.54
0.1539	0.7273	6	0.1548	17.11	17.11	20.54
0.1521	0.9697	8	0.1480	17.11	17.11	20.54

Framework versions

PEFT 0.17.1
Transformers 4.56.1
Pytorch 2.7.1+cu126
Datasets 4.0.0
Tokenizers 0.22.1

Tibogoss/Qwen3-8B-test

workspace/outputs/FT_v2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Tibogoss/
Qwen3-8B-test