Safetensors
qwen2
max_seq_length=16384
lr=2e-5
batch_size=1
grad_accum=16
epochs=3
qlora
quantize=4bit_nf4
lora_rank=64
lora_alpha=128
lora_dropout=0.05
completion_only_loss
eval_size=0.1
cosine_schedule
warmup=0.05
bf16
run_name=sft-v4.5
hub_model_id=tinyllms/qwen2.5-7b-instruct-sft-aime
ray_job=raysubmit_iVkMK5RLqcXEJPeA
File size: 133 Bytes
2562da8 | 1 2 3 4 | version https://git-lfs.github.com/spec/v1
oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
size 11421896
|