See axolotl config

axolotl version: 0.4.1

base_model: Qwen/Qwen2.5-3B-Instruct

strict: false

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

chat_template: chatml
datasets:
  - path: AlekseyKorshuk/ai-detection-booksum-complete-cleaned-human-ai-sft
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
    roles:
      user:
        - user
      assistant:
        - assistant

val_set_size: 0.05
output_dir: ./outputs/out

eval_table_size: 0
eval_max_new_tokens: 1

sequence_len: 8192
sample_packing: false
eval_sample_packing: false
pad_to_sequence_len: false

wandb_project: ai-seo-rewriter
wandb_entity:
wandb_watch:
wandb_name: ai-detection-booksum-complete-cleaned-human-ai-sft-qwen-3b
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 16
eval_batch_size: 16
num_epochs: 1
optimizer: adamw_torch
# adam_beta1: 0.9
# adam_beta2: 0.95
max_grad_norm: 1.0
# adam_epsilon: 0.00001
lr_scheduler: cosine
learning_rate: 8e-6

train_on_inputs: false
group_by_length: false
bf16: true
fp16: false
tf32: true

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

warmup_ratio: 0.1
evals_per_epoch: 10
saves_per_epoch: 1
debug:
deepspeed:
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: Qwen2DecoderLayer
  fsdp_state_dict_type: FULL_STATE_DICT



hub_model_id: AlekseyKorshuk/ai-detection-booksum-complete-cleaned-human-ai-sft-qwen-3b

ai-detection-booksum-complete-cleaned-human-ai-sft-qwen-3b

This model is a fine-tuned version of Qwen/Qwen2.5-3B-Instruct on the None dataset. It achieves the following results on the evaluation set:

Loss: 1.0386

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: 8e-06
train_batch_size: 16
eval_batch_size: 16
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 128
total_eval_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 40
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
2.3436	0.0025	1	2.4016
1.2487	0.1015	41	1.1937
1.1094	0.2030	82	1.1329
1.1286	0.3045	123	1.1040
1.0279	0.4059	164	1.0828
0.9941	0.5074	205	1.0676
0.9925	0.6089	246	1.0553
0.9619	0.7104	287	1.0458
0.9993	0.8119	328	1.0408
0.9544	0.9134	369	1.0386