unfiltered-1.3b-grpo-math-expert

unfiltered-1.3b-grpo-math-expert

1.36B fully-unbounded chat model post-trained with GRPO on top of SFT.

Part of the LittleLearner scale-up study (pedagogically-controlled knowledge exposure): Qwen3 dense LMs trained on a corpus filtered to U.S. K-5 material (bounded) vs an unfiltered FineWeb-Edu corpus (unbounded), to measure what an interpretable knowledge boundary costs and grants.

Note: This checkpoint was post-trained with GRPO on mathematical reasoning tasks to probe achievable performance on MathCAMPS. As a result, its behavior is specialized toward mathematical reasoning and may not preserve general-purpose chat capabilities; responses may also exhibit a tendency toward math-oriented reasoning or output.

Model

• Architecture: Qwen3 dense (Qwen3ForCausalLM).
• Size: 1.358B params, hidden 2048, 26 layers, 16 query / 8 KV heads, FFN 5632. Context: 4096.
• Tokenizer: custom 64k byte-level BPE with per-digit splitting (ChatML special tokens).
• Pretraining: 88B tokens on unfiltered FineWeb-Edu (score >= 2, no grade filter). WSD schedule, sharded Muon, MXFP8, Megatron-Core on 8xB200.
• SFT: supervised fine-tuned on unbounded chat data (lr 1e-5, 3 epochs).
• RL (GRPO): segmented policy re-banding on a verifiable-answer pool: temperature-1.0 segments to a plateau, then one segment at temperature 1.3 (a one-shot unlock).

Evaluation

MathCAMPS:

• K-5 pass@64 70.6 / pass@1 45.7
• beyond-K-5 pass@64 45.6 / pass@1 16.5

Usage

# transformers (chat)
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "manueldeprada/littlelearner-1.3b-unbounded-grpo"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo, dtype="bfloat16", device_map="cuda")
msgs = [{"role": "user", "content": "Liam has 3 apples and buys 4 more. How many apples does he have?"}]
ids = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt").to(model.device)
out = model.generate(ids)
print(tok.decode(out[0, ids.shape[1]:], skip_special_tokens=True))

# vLLM
from vllm import LLM
repo = "manueldeprada/littlelearner-1.3b-unbounded-grpo"
llm = LLM(repo)
msgs = [{"role": "user", "content": "Liam has 3 apples and buys 4 more. How many apples does he have?"}]
print(llm.chat(msgs)[0].outputs[0].text)