See axolotl config

axolotl version: 0.6.0

base_model: meta-llama/Llama-3.2-1B
hub_model_id: minpeter/Llama-3.2-1B-Instruct-chatml

load_in_8bit: false
load_in_4bit: false
strict: false

chat_template: chatml
datasets:
  - path: philschmid/guanaco-sharegpt-style
    type: chat_template
    field_messages: conversations
    message_field_role: from
    message_field_content: value
  - path: teknium/OpenHermes-2.5
    type: chat_template
    field_messages: conversations
    message_field_role: from
    message_field_content: value
dataset_prepared_path: last_run_prepared
val_set_size: 0.05

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

wandb_project: "axolotl"
wandb_entity: "kasfiekfs-e"
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

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

warmup_steps: 100
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>
  eos_token: <|im_end|>
tokens:
  - "<|im_start|>"

Llama-3.2-1B-Instruct-chatml

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on the philschmid/guanaco-sharegpt-style and the teknium/OpenHermes-2.5 datasets. It achieves the following results on the evaluation set:

• Loss: 0.8542

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• total_eval_batch_size: 2
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT 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: 100
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.1381	0.0003	1	1.1334
0.8563	0.5	1466	0.8594
0.8282	1.0	2932	0.8542

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0