mariiazhiv/CyTHIA-Mixtral-8x7B

See axolotl config

axolotl version: 0.8.0.dev0

base_model: mistralai/Mixtral-8x7B-Instruct-v0.1
datasets:
  - path: mariiazhiv/cybersecurity_qa
    type: alpaca
    split: train
  - path: mariiazhiv/cybersecurity_qa
    type: alpaca
    split: validation
dataset_prepared_path: last_run_prepared
output_dir: ./outputs/mymodel

sequence_len: 1024
adapter: lora

lora_r: 16
lora_alpha: 32
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - v_proj
  - k_proj
  - o_proj
  - gate_proj
  - down_proj
  - up_proj

gradient_accumulation_steps: 8
micro_batch_size: 4
num_epochs: 3
optimizer: adamw_bnb_8bit
learning_rate: 0.00002
load_in_8bit: false
train_on_inputs: false
bf16: true
fp16: false
gradient_checkpointing: true

eval_steps: 50
save_steps: 50
logging_steps: 10

special_tokens:
  pad_token: "<|pad|>"

outputs/mymodel

This model is a fine-tuned version of mistralai/Mixtral-8x7B-Instruct-v0.1 on the mariiazhiv/cybersecurity_qa and the mariiazhiv/cybersecurity_qa 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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_BNB 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: 2
• num_epochs: 3.0

Training results

Framework versions

• PEFT 0.14.0
• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0