github/gpt4all
1
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2025-08-04 09:32:29 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: full model config

Browse Source
This commit is contained in:
Zach Nussbaum 2023-03-27 17:30:57 +00:00
parent 1a95f68494
commit 80227dc164
1 changed files with 10 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
configs/train/finetune.yaml
View File

@ -2,27 +2,29 @@
model_name: "zpn/llama-7b" model_name: "zpn/llama-7b"
tokenizer_name: "zpn/llama-7b" tokenizer_name: "zpn/llama-7b"
gradient_checkpointing: true gradient_checkpointing: true
save_name: "nomic-ai/vicuna-full-multi-turn"
# dataset # dataset
streaming: false streaming: false
num_proc: 64 num_proc: 64
dataset_path: "data.jsonl" dataset_path: "data_multiturn"
max_length: 512 max_length: 1024
batch_size: 32 batch_size: 32
# train dynamics # train dynamics
lr: 5.0e-5 lr: 5.0e-5
eval_every: 2000 eval_every: 800
eval_steps: 100 eval_steps: 100
save_every: 2000 save_every: 800
output_dir: "ckpts/llama-7b" output_dir: "ckpts/llama-7b-full-multi"
checkpoint: null checkpoint: null
lora: false lora: false
warmup_steps: 100 warmup_steps: 100
num_epochs: 2
# logging # logging
wandb: false wandb: true
wandb_entity: zanussbaum wandb_entity: vicuna
wandb_project: llama wandb_project_name: vicuna
seed: 42 seed: 42