Modelpublic

cgifbribcgfbi/Llama-3.3-70B-Instruct-abliterated-finetuned-chem-claude-5-comp3-sort-comp

sourceHugging Facellama3.3updated 1y agoView on Hugging Face

0likes7downloads

Model Card

<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/> <details><summary>See axolotl config</summary>

axolotl version: 0.9.1

yaml

base_model: huihui-ai/Llama-3.3-70B-Instruct-abliterated-finetuned
load_in_8bit: false
load_in_4bit: true
adapter: qlora
wandb_name: Llama-3.3-70B-Instruct-abliterated-finetuned-chem-claude-5-comp3-sort-comp
output_dir: ./outputs/out/Llama-3.3-70B-Instruct-abliterated-finetuned-chem-claude-5-comp3-sort-comp
hub_model_id: cgifbribcgfbi/Llama-3.3-70B-Instruct-abliterated-finetuned-chem-claude-5-comp3-sort-comp

tokenizer_type: AutoTokenizer
push_dataset_to_hub:
strict: false

datasets:
  - path: dset_comp3.0_sortcomplexity_pat400_in5_num5000_5000.jsonl
    type: chat_template
    field_messages: messages

dataset_prepared_path: last_run_prepared
val_set_size: 0.04
save_safetensors: true

sequence_len: 2700
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: LlamaDecoderLayer
  fsdp_state_dict_type: FULL_STATE_DICT
  fsdp_sharding_strategy: FULL_SHARD
special_tokens:
  pad_token: <|finetune_right_pad_id|>

</details><br>

Llama-3.3-70B-Instruct-abliterated-finetuned-chem-claude-5-comp3-sort-comp

This model is a fine-tuned version of huihui-ai/Llama-3.3-70B-Instruct-abliterated-finetuned on the dsetcomp3.0sortcomplexitypat400in5num50005000.jsonl dataset. It achieves the following results on the evaluation set:

—Loss: 0.3517

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
—trainbatchsize: 4
—evalbatchsize: 4
—seed: 42
—distributed_type: multi-GPU
—num_devices: 4
—totaltrainbatch_size: 16
—totalevalbatch_size: 16
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 10
—num_epochs: 4.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.5023	0.0060	1	0.6780
0.4863	0.3333	56	0.4702
0.4542	0.6667	112	0.4127
0.427	1.0	168	0.3917
0.3873	1.3333	224	0.3772
0.3948	1.6667	280	0.3686
0.3857	2.0	336	0.3650
0.3591	2.3333	392	0.3583
0.368	2.6667	448	0.3558
0.3644	3.0	504	0.3549
0.3448	3.3333	560	0.3521
0.3568	3.6667	616	0.3518
0.3601	4.0	672	0.3517

Framework versions

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