# === Model Configuration ===
base_model: unsloth/Llama-3.3-70B-Instruct
load_in_8bit: false
load_in_4bit: true

# === Training Setup ===
num_epochs: 2
micro_batch_size: 2
gradient_accumulation_steps: 1
sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

# === Hyperparameter Configuration ===
optimizer: came_pytorch
learning_rate: 1e-5
lr_scheduler: rex
weight_decay: 0.01
warmup_steps: 0
cosine_min_lr_ratio: 0.1

# === LoRA Configuration ===
adapter: qlora
lora_r: 16
lora_alpha: 32
lora_dropout: 0.25
lora_target_modules:
lora_target_linear: true

# === Data Configuration ===
chat_template: llama3
datasets:
  - path: allura-org/inkmix-v2.1
    type: chat_template
    split: train
    field_messages: conversations
    message_field_role: from
    message_field_content: value

dataset_prepared_path: last_run_prepared

# === Plugins ===
plugins:
  - axolotl.integrations.liger.LigerPlugin
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin

# === Hardware Optimization ===
gradient_checkpointing: unsloth
gradient_checkpointing_kwargs:
  use_reentrant: false
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
cut_cross_entropy: true
deepspeed: deepspeed_configs/zero3_bf16.json

# === Wandb Tracking ===
# wandb_project: [WANDB_PROJECT_NAME]
# wandb_entity: [WANDB_ENTITY]
# wandb_name: [WANDB_RUN_NAME]

# === MLflow Tracking ===
mlflow_tracking_uri: https://public-tracking.mlflow-e00zzfjq11ky6jcgtv.backbone-e00bgn6e63256prmhq.msp.eu-north1.nebius.cloud
mlflow_experiment_name: l3.3-70b-inkmixv2.1-qlora
# hf_mlflow_log_artifacts: true
# Disable logging artifacts to mlflow if model is big (it hangs):
hf_mlflow_log_artifacts: false

# === Checkpointing ===
saves_per_epoch: 1
save_total_limit: 2

# === Advanced Settings ===
output_dir: ./ckpts
bf16: auto
flash_attention: true
train_on_inputs: false
group_by_length: false
logging_steps: 1
special_tokens:
  pad_token: <|finetune_right_pad_id|>
  eos_token: <|eot_id|>

ckpts

This model is a fine-tuned version of unsloth/Llama-3.3-70B-Instruct on the allura-org/inkmix-v2.1 dataset.

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: 1e-05
train_batch_size: 2
eval_batch_size: 2
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 16
total_eval_batch_size: 16
optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
num_epochs: 2.0

Training results

Framework versions

PEFT 0.14.0
Transformers 4.50.0.dev0
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.1

estrogen/Bigger-Body-70b-ep2-adpt

ckpts

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

estrogen/
Bigger-Body-70b-ep2-adpt