library_name: peft license: gemma base_model: mlabonne/gemma-3-27b-it-abliterated tags:

• axolotl
• generated_from_trainer datasets:
• dset_comp3.0_sortpatent_count_pat400_in5_5000.jsonl model-index:
• name: gemma-3-27b-it-abliterated-chem-claude-5-comp3-sort-pate results: []

axolotl version: 0.9.1

base_model: mlabonne/gemma-3-27b-it-abliterated
load_in_8bit: false
load_in_4bit: true
adapter: qlora
wandb_name: gemma-3-27b-it-abliterated-chem-claude-5-comp3-sort-pate
output_dir: ./outputs/out/gemma-3-27b-it-abliterated-chem-claude-5-comp3-sort-pate
hub_model_id: cgifbribcgfbi/gemma-3-27b-it-abliterated-chem-claude-5-comp3-sort-pate

tokenizer_type: AutoTokenizer
push_dataset_to_hub:
strict: false

datasets:
  - path: dset_comp3.0_sortpatent_count_pat400_in5_5000.jsonl
    type: chat_template
    field_messages: messages

dataset_prepared_path: last_run_prepared
val_set_size: 0.04
save_safetensors: true

sequence_len: 2833
sample_packing: true
pad_to_sequence_len: true

lora_r: 64
lora_alpha: 32
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true

wandb_mode:
wandb_project: finetune-sweep
wandb_entity: gpoisjgqetpadsfke
wandb_watch:
wandb_run_id:
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 4  # This will be automatically adjusted based on available GPU memory
num_epochs: 4
optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 0.00002

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

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

warmup_steps: 10
evals_per_epoch: 3
saves_per_epoch: 1
weight_decay: 0.01
fsdp:
  - full_shard
  - auto_wrap
fsdp_config:
  fsdp_limit_all_gathers: true
  fsdp_sync_module_states: true
  fsdp_offload_params: false
  fsdp_use_orig_params: false
  fsdp_cpu_ram_efficient_loading: true
  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
  fsdp_transformer_layer_cls_to_wrap: Gemma3DecoderLayer
  fsdp_state_dict_type: FULL_STATE_DICT
  fsdp_sharding_strategy: FULL_SHARD
special_tokens:
  pad_token: <|finetune_right_pad_id|>

gemma-3-27b-it-abliterated-chem-claude-5-comp3-sort-pate

This model is a fine-tuned version of mlabonne/gemma-3-27b-it-abliterated on the dset_comp3.0_sortpatent_count_pat400_in5_5000.jsonl dataset. It achieves the following results on the evaluation set:

• Loss: 0.3741

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 16
• total_eval_batch_size: 16
• 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
• lr_scheduler_warmup_steps: 10
• num_epochs: 4.0

Training results

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1