58 lines
1.2 KiB
YAML
58 lines
1.2 KiB
YAML
# EBFT Strided: Full-parameter Llama-3.1-8B on SwallowCode, 100 steps
|
|
# Feature network is CPU-offloaded to fit in single 32GB GPU
|
|
#
|
|
# Run: CUDA_VISIBLE_DEVICES=0 python -m axolotl.cli.train examples/ebft/llama-8b-ebft-strided-fft.yaml
|
|
|
|
base_model: meta-llama/Llama-3.1-8B
|
|
rl: ebft
|
|
|
|
ebft:
|
|
mode: strided
|
|
stride: 8
|
|
context_length: 8
|
|
generate_max_len: 8
|
|
n_samples_per_prompt: 4
|
|
temperature: 0.6
|
|
top_p: 1.0
|
|
feature_layers: [0.25, 0.5, 0.75]
|
|
embed_method: last_token
|
|
use_whitening: true
|
|
alignment_coef: 1.0
|
|
diversity_coef: 1.0
|
|
rl_coef: 1.0
|
|
ce_coef: 0.0
|
|
advantage_estimator: rloo
|
|
|
|
datasets:
|
|
- path: sjelassi/swallow_code_20m
|
|
type: ebft_pretrain.transform
|
|
split: train[:5000]
|
|
|
|
sequence_len: 1024
|
|
micro_batch_size: 1
|
|
gradient_accumulation_steps: 4
|
|
num_epochs: 1
|
|
max_steps: 100
|
|
|
|
learning_rate: 1.0e-6
|
|
optimizer: adamw_torch_fused
|
|
lr_scheduler: cosine
|
|
warmup_steps: 10
|
|
weight_decay: 0.01
|
|
|
|
bf16: auto
|
|
gradient_checkpointing: true
|
|
gradient_checkpointing_kwargs:
|
|
use_reentrant: false
|
|
|
|
special_tokens:
|
|
pad_token: "<|end_of_text|>"
|
|
|
|
val_set_size: 0.0
|
|
output_dir: ./outputs/ebft-llama8b-strided
|
|
|
|
wandb_project: ebft
|
|
wandb_name: llama8b-strided-fft-100steps
|
|
logging_steps: 1
|
|
save_steps: 50
|