Training
Table | |
|---|
| Base model | google/gemma-4-E2B-it |
| Algorithm | GRPO (leave-one-out baseline, group reward normalization, token-level loss). Responses that hit the budget are truncated and scored as-is. |
Generation budget (max_new_tokens) | 1,024 |
| Data | DeepScaleR (math), 3 epochs max |
| Batch | 32 prompts × 8 rollouts per step |
| Optimizer | Adam, cosine LR schedule, peak LR 3e-6, 10 warmup steps |
| Steps | 300 |
| Reward | binary answer correctness (\boxed{} extraction) |
| Weights dtype | F32 |
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
from transformers import AutoModelForCausalLM, AutoTokenizer repo = "budget-internalization-iclr2027/gemma4-e2b-1k-grpo-luckymarmot-s300"tok = AutoTokenizer.from_pretrained(repo)model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="auto", device_map="auto")
With vLLM: vllm serve budget-internalization-iclr2027/gemma4-e2b-1k-grpo-luckymarmot-s300
License
Inherits the license of the base model (google/gemma-4-E2B-it).