ModalityDance/slpo-codi-gpt2

CODI + SLPO (GPT-2)

Surrogate Latent Policy Optimization (SLPO) checkpoint on top of CODI (GPT-2 124M). This is the CODI+SLPO model reported in the paper SLPO: Scaling Latent Reasoning with Surrogate Policy Optimization.

Model Details

• Backbone: CODI / GPT-2 (ModalityDance/latent-tts-codi)
• Method: stopping-gate cold start → SLPO (RLOO) with adaptive latent stopping
• Special tokens: <|latent|>, <|start-latent|>, <|end-latent|>
• Recommended gate threshold: 0.7
• Max latent length: 12

Results (paper main table, Acc)

Deterministic accuracy with dropout disabled and learned stop gate:

Benchmark	Acc	Mean latent length
GSM8K	42.76	11.83
GSM-Hard	9.71	11.94
MultiArith	90.52	11.44

Related

• Paper (arXiv): coming soon — placeholder https://arxiv.org/abs/XXXX.XXXXX
• Hugging Face Paper: coming soon — placeholder https://huggingface.co/papers/XXXX.XXXXX
• Code: ModalityDance/SLPO
• Project page: modalitydance.github.io/SLPO
• Base model: ModalityDance/latent-tts-codi
• Sibling: ModalityDance/slpo-coconut-gpt2
• Collection: ModalityDance/SLPO

Installation

git clone https://github.com/ModalityDance/SLPO.git
cd SLPO
pip install -r requirements.txt   # plus a CUDA PyTorch build
hf download ModalityDance/slpo-codi-gpt2 --local-dir checkpoints/slpo-codi-gpt2

Quick Start

Batched eval (paper Acc settings):

CKPT=checkpoints/slpo-codi-gpt2 \
MODEL_TYPE=codi STOP_POLICY=gate \
STOP_GATE_THRESHOLD=0.7 MAX_LATENT_LENGTH=12 \
DATA=data/gsm_test.json \
bash scripts/eval.sh

Minimal Python (from the repo root; needs the SLPO latent generation stack):

import torch
from transformers import AutoTokenizer

from src.models.generation import LatentGenerationMixin, LatentGenerationConfig
from src.paths import get_model_class

model_id = "ModalityDance/slpo-codi-gpt2"
backbone_cls = get_model_class("codi")

class LatentModel(backbone_cls, LatentGenerationMixin):
    pass

tokenizer = AutoTokenizer.from_pretrained(model_id)
if tokenizer.pad_token is None:
    tokenizer.pad_token = tokenizer.eos_token

model = LatentModel.from_pretrained(model_id)
model.eval()

question = (
    "Janet's ducks lay 16 eggs per day. She eats three for breakfast every morning "
    "and bakes muffins for her friends every day with four. She sells the remainder "
    "at the farmers' market daily for $2 per fresh duck egg. "
    "How much in dollars does she make every day at the farmers' market?"
)
prompt = question + "<|start-latent|>"
inputs = tokenizer(prompt, return_tensors="pt")

gen_cfg = LatentGenerationConfig(
    stop_policy="gate",
    max_latent_length=12,
    stop_gate_threshold=0.7,
    max_new_tokens=128,
    pad_token_id=tokenizer.pad_token_id,
    eos_token_id=tokenizer.eos_token_id,
    bos_token_id=tokenizer.bos_token_id,
)

with torch.no_grad():
    output = model.generate(**inputs, generation_config=gen_cfg)
sequences = output.sequences if hasattr(output, "sequences") else output
print(tokenizer.decode(sequences[0], skip_special_tokens=True))

Citation

@misc{you2026slpo,
  title   = {SLPO: Scaling Latent Reasoning with Surrogate Policy Optimization},
  author  = {You, Runyang and Liu, Zhiyuan and Li, Yongqi and Li, Wenjie},
  year    = {2026},
  note    = {Code: https://github.com/ModalityDance/SLPO}
}