base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
trust_remote_code: true
load_in_8bit: true
bf16: true

adapter: lora
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - v_proj

pretraining_dataset:
  - name: default
    path: aegis-nvim/neovim-help
    type: pretrain
    text_column: text
    trust_remote_code: false

val_set_size: 0
sequence_len: 2048
group_by_length: false
max_steps: 5000

micro_batch_size: 16
gradient_accumulation_steps: 2
gradient_checkpointing: true
optimizer: adamw_bnb_8bit
learning_rate: 0.0001
lr_scheduler: cosine
warmup_ratio: 0.03
max_grad_norm: 1.0

output_dir: ./outputs/dapt-lora
save_strategy: steps
save_steps: 5000
save_total_limit: 1
save_safetensors: true

outputs/dapt-lora

This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 on an unknown 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: 0.0001
train_batch_size: 16
eval_batch_size: 16
seed: 42
gradient_accumulation_steps: 2
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: 150
training_steps: 5000

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

aegis-nvim/neovim-help-adapter

outputs/dapt-lora

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

aegis-nvim/
neovim-help-adapter