# === Model Configuration ===
base_model: Columbidae/mixed-model-prune-52
load_in_8bit: false
load_in_4bit: true

# === HF Configuration === 
hub_model_id: Columbidae/mixed-model-prune-trained-ws
hub_strategy: "every_save"

# === Training Setup ===
num_epochs: 1
micro_batch_size: 1
#eval_batch_size: 1
gradient_accumulation_steps: 4
sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true

# === Evaluation ===
#val_set_size: 100
eval_strategy: "no"
#evals_per_epoch: 10
eval_table_size:
eval_max_new_tokens: 256
eval_sample_packing: true

# === LoRA Configuration ===
adapter: qlora
lora_model_dir:
lora_r: 64
lora_alpha: 32
lora_dropout: 0.5
lora_target_linear: 
lora_fan_in_fan_out:
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

# === Hyperparameter Configuration ===
optimizer: paged_ademamix_8bit #apollo_adamw
# Apollo-mini configuration:
#optim_args: "proj=random,rank=1,scale=128.0,scale_type=tensor,update_proj_gap=200"
# Regular Apollo configuration:
# optim_args: 
#optim_target_modules: all_linear
learning_rate: 1e-5
lr_scheduler: cosine
weight_decay: 0.01
warmup_ratio: 0.05


# === Data Configuration ===
datasets:
  - path: Columbidae/merge-glue-4k
    data_files: conversation-glue-4k.json
    type: chat_template
    split: train
    field_messages: conversations
    message_field_role: from
    message_field_content: value
  - path: Columbidae/merge-glue-4k
    data_files: completion-glue-4k.json
    type: completion
    split: train
    field: text

dataset_prepared_path: last_run_prepared
chat_template: tokenizer_default
# Example custom template:
# chat_template: jinja
# chat_template_jinja: |
#   {{- bos_token }}{%- for message in messages %}
#   {%- if message['role'] == 'system' %}
#   {{- '[SYSTEM_PROMPT]' + message['content'] + '[/SYSTEM_PROMPT]' }}
#   {%- elif message['role'] == 'user' %}
#   {{- '[INST]' + message['content'] + '[/INST]' }}
#   {%- elif message['role'] == 'assistant' %}
#   {{- message['content'] + eos_token }}
#   {%- endif %}
#   {%- endfor %}

# === 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
liger_fused_linear_cross_entropy: true
unsloth_cross_entropy_loss: true
#cut_cross_entropy: true
# Only if using multiple GPUs:
deepspeed: axolotl/deepspeed_configs/zero3_bf16.json

# === Wandb Tracking ===
wandb_project: Qwen-27
# wandb_entity: [WANDB_ENTITY]
# wandb_name: [WANDB_RUN_NAME]

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

# === Advanced Settings ===
output_dir: ./ckpts
bf16: auto
flash_attention: true
train_on_inputs: false
group_by_length: false
save_safetensors: true
logging_steps: 1
gc_steps: 10
seed: 69

mixed-model-prune-trained-ws

This model is a fine-tuned version of Columbidae/mixed-model-prune-52 on the Columbidae/merge-glue-4k and the Columbidae/merge-glue-4k datasets.

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: 1
eval_batch_size: 1
seed: 69
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 4
total_train_batch_size: 16
total_eval_batch_size: 4
optimizer: Use OptimizerNames.PAGED_ADEMAMIX_8BIT and the args are: No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 11
num_epochs: 1.0

Training results

Framework versions

PEFT 0.14.0
Transformers 4.48.3
Pytorch 2.5.1+cu124
Datasets 3.2.0
Tokenizers 0.21.0