CoolFace
Modelpublic

budget-internalization-iclr2027/qwen3.5-4b-16k-sft-meetwombat-s79

sourceHugging Faceapache-2.0updated 2d agoView on Hugging Face
0likes15downloads
Model Card

Qwen3.5-4B · SFT on 10k-16k RFT data · meetwombat

Supervised finetune of Qwen/Qwen3.5-4B on its own reward-filtered solutions to DeepScaleR math problems (rejection-sampling finetuning), restricted to long responses of 10k-16k tokens. Released as part of an anonymous ICLR 2027 submission.

Run ID (petname): `meetwombat` · checkpoint step 79

Training

Base modelQwen/Qwen3.5-4B
AlgorithmSFT (next-token loss on the response)
Dataself-generated, correctness-filtered Qwen3.5-4B solutions to DeepScaleR problems; subset 10k-16k (long responses of 10k-16k tokens), one solution per problem
Max sequence length18,432
Batch32 sequences per step
OptimizerAdam, cosine LR schedule, peak LR 1e-06
Epochs1
Steps79
Weights dtypeBF16

Training prompt (user turn, rendered with the base model's chat template):

Think step-by-step to solve the following problem. Output your answer inside of \\boxed{} tags.:
{problem}

Let's think step-by-step

Usage

python
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "budget-internalization-iclr2027/qwen3.5-4b-16k-sft-meetwombat-s79"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="auto", device_map="auto")

With vLLM: vllm serve budget-internalization-iclr2027/qwen3.5-4b-16k-sft-meetwombat-s79

License

Inherits the license of the base model (Qwen/Qwen3.5-4B).