@HuggingPapers: LoopCoder-v2 is out A 7B model trained on 18T tokens that scores 64.4 on SWE-bench Verified with just two loops, beatin…
Summary
LoopCoder-v2, a 7B model trained on 18T tokens, achieves 64.4 on SWE-bench Verified with only two loops, outperforming models 30x larger. The model and code are available on Hugging Face.
View Cached Full Text
Cached at: 06/17/26, 01:54 PM
LoopCoder-v2 is out
A 7B model trained on 18T tokens that scores 64.4 on SWE-bench Verified with just two loops, beating models 30x larger.
Adding a third loop makes it worse.
Model and code are on Hugging Face. https://t.co/nyHlt7suMB
Similar Articles
20B Looping model (paper) matches or beats Qwen3 Coder 30B at 10% of pre-training tokens
Loopie models use a looped transformer architecture to match or exceed Qwen3 Coder 30B performance with only 10% of the pre-training tokens, demonstrating strong reasoning abilities and efficient scaling.
Multilingual-Multimodal-NLP/LoopCoder-V2 · Hugging Face
LoopCoder-V2 is a 7B instruction-tuned code model built on the Parallel Loop Transformer (PLT), demonstrating non-monotonic test-time scaling with two loops providing the best gain-cost trade-off and significant improvements over baselines on code generation and reasoning benchmarks.
@DorothyDDU: LoopCoder-v2 is out Loop Transformers reuse the same block for recurrent hidden-state refinement — letting models “thin…
This paper introduces LoopCoder-v2, a family of 7B parameter parallel loop transformers for code generation, and studies the optimal number of loops, finding that two loops yield significant gains while more loops cause degradation.
@arXivBangers: Closing the Loop: Practical Training Recipes for Looped Language Models New training recipes enable looped language mod…
The paper introduces practical training recipes that cut looped (recurrent) language model training from 7.7T to 310B tokens, with a 1.4B LoopLM outperforming a parameter-matched dense baseline on all 12 benchmarks, and shows how to convert pretrained dense models like Qwen3-1.7B-Base into looped ones with minimal added parameters.
@ClementDelangue: Kog open-sourced on @huggingface the 2B model that they used to show a model running at 3,000+ tokens per second. Very …
Kog has open-sourced the Laneformer 2B model, a 2.3B parameter instruction-tuned coding model designed for high-speed decoding, achieving over 3,000 tokens per second by prioritizing latency from the architecture stage.