Helios Reasoner 7B Prod | Sweet Tea Studio
Resources / Helios Reasoner 7B Prod Helios Reasoner 7B Prod Helios-Reasoner-7B is our flagship open reasoning model. Through extended RLHF and a curated math/code post-training mix, it now closes much of the gap with frontier closed models on hard multi-step reasoning. Evaluation Results Comprehensive Benchmark Results
Kind feature-extraction Version vaf10bea62f0a9d76c157d20894ec6de7665245ab License apache-2.0 Publisher Procedure2012 D grade
Model source
Kind feature-extraction
Version vaf10bea62f0a9d76c157d20894ec6de7665245ab
License apache-2.0
Publisher Procedure2012
Source Hugging Face
Helios-Reasoner-7B ## 1. Introduction Helios-Reasoner-7B is our flagship open reasoning model. Through extended RLHF and a curated math/code post-training mix, it now closes much of the gap with frontier closed models on hard multi-step reasoning. ## 2. Evaluation Results ### Comprehensive Benchmark Results | | Benchmark | Falcon-S | Mistral-Lite | Helios-v1 | Helios-Reasoner-7B | |---|---|---|---|---|---| | Source excerpts
3 excerpts Helios-Reasoner-7B is our flagship open reasoning model. Through extended RLHF and a curated math/code post-training mix, it now closes much of the gap with frontier closed models on hard multi-step reasoning. Evaluation Results Comprehensive Benchmark Results
Core Reasoning Tasks
Language Understanding
Generation Tasks
Specialized Capabilities
research@helios-labs.ai
Readme excerpt
Jul 13
Helios-Reasoner-7B is our flagship open reasoning model. Through extended RLHF and a curated math/code post-training mix, it now closes much of the gap with frontier closed models on hard multi-step reasoning.
procedure2012/Helios-Reasoner-7B-Prod
