SWE-rebench-V2 is a curated dataset of software-engineering tasks derived from real GitHub issues and pull requests.
Nebius
company
Verified
AI & ML interests
AI-centric cloud platform ready for intensive workloads Training-ready platform with NVIDIA® H100 Tensor Core GPUs. Competitive pricing. Dedicated support.
Recent Activity
View all activity
Papers
SlimSpec: Low-Rank Draft LM-Head for Accelerated Speculative Decoding
LK Losses: Direct Acceptance Rate Optimization for Speculative Decoding
Training data for speculative decoding draft models, containing model-generated responses for 660K prompts from Infinity-Instruct-0625.
-
nebius/Llama-3.1-8B-Instruct-Infinity-Instruct-0625
Viewer • Updated • 660k • 397 -
nebius/Llama-3.3-70B-Instruct-Infinity-Instruct-0625
Viewer • Updated • 660k • 21 -
nebius/gpt-oss-20b-Infinity-Instruct-0625
Viewer • Updated • 660k • 609 -
nebius/gpt-oss-120b-Infinity-Instruct-0625
Viewer • Updated • 659k • 60
Dataset of 67,074 OpenHands trajectories collected with Qwen3-Coder-480B-A35B-Instruct and two RFT checkpoints trained on the data
SWE-rebench-V2 is a curated dataset of software-engineering tasks derived from real GitHub issues and pull requests.
Training data for speculative decoding draft models, containing model-generated responses for 660K prompts from Infinity-Instruct-0625.
-
nebius/Llama-3.1-8B-Instruct-Infinity-Instruct-0625
Viewer • Updated • 660k • 397 -
nebius/Llama-3.3-70B-Instruct-Infinity-Instruct-0625
Viewer • Updated • 660k • 21 -
nebius/gpt-oss-20b-Infinity-Instruct-0625
Viewer • Updated • 660k • 609 -
nebius/gpt-oss-120b-Infinity-Instruct-0625
Viewer • Updated • 659k • 60
High-performance speculative decoding draft models trained using LK losses, a novel training objectives that directly optimize acceptance rate
Dataset of 67,074 OpenHands trajectories collected with Qwen3-Coder-480B-A35B-Instruct and two RFT checkpoints trained on the data