adapter: lora
base_model: mistralai/Mistral-Nemo-Instruct-2407
bf16: auto
datasets:
- path: quentintousart/vian-hierarchizer-training
  data_files:
    - training_data.jsonl
  type: chat_template
  ds_type: json
chat_template: mistral_v2v3
special_tokens:
  pad_token: "<pad>"
gradient_accumulation_steps: 4
learning_rate: 0.0002
load_in_8bit: true
lora_alpha: 32
lora_dropout: 0.05
lora_r: 16
lora_target_modules:
- q_proj
- v_proj
- k_proj
- o_proj
- gate_proj
- down_proj
- up_proj
micro_batch_size: 2
num_epochs: 3
optimizer: adamw_bnb_8bit
output_dir: ./outputs/vian-nemo
sequence_len: 4096
train_on_inputs: false
hub_model_id: quentintousart/mistral-nemo-vian-hierarchizer
hub_strategy: end
hf_use_auth_token: true
save_safetensors: true
logging_steps: 10
warmup_steps: 10

mistral-nemo-vian-hierarchizer

This model is a fine-tuned version of mistralai/Mistral-Nemo-Instruct-2407 on the quentintousart/vian-hierarchizer-training 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: 0.0002
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
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: 10
training_steps: 225

Training results

Framework versions

PEFT 0.17.1
Transformers 4.57.0
Pytorch 2.7.1+cu126
Datasets 4.0.0
Tokenizers 0.22.1