Steffen Röcker PRO

sroecker's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

upvoted a collection about 5 hours ago

Ternary Bonsai

Collection

1.58-bit Bonsai models • 5 items • Updated about 6 hours ago • 12

upvoted an article about 6 hours ago

view article

Article

The PR you would have opened yourself

• 
• 

about 23 hours ago

•

32

upvoted a paper about 13 hours ago

BERT-as-a-Judge: A Robust Alternative to Lexical Methods for Efficient Reference-Based LLM Evaluation

Paper • 2604.09497 • Published 7 days ago • 26

upvoted 2 collections 1 day ago

Parcae

Collection

A family of stable looped models • 4 items • Updated 2 days ago • 3

HLWQ Unified (Weights Q5 + KV Cache Q3)

Collection

Full-stack HLWQ: Q5 weights + torchao INT4 + Q3 KV cache · formerly PolarQuant Unified • 17 items • Updated 4 days ago • 3

upvoted a collection 2 days ago

Rethink_SFT_generalization

Collection

Repo for paper Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capability. • 40 items • Updated 5 days ago • 16

upvoted an article 3 days ago

view article

Article

How we OCR'ed 30,000 papers using Codex, open OCR models and Jobs

• 

9 days ago

•

53

upvoted a changelog 9 days ago

view changelog

Hugging Face Changelog

Agent Traces on the Hub

9 days ago

• 100

upvoted a collection 9 days ago

GLM-5.1

Collection

2 items • Updated 9 days ago • 58

upvoted a collection 14 days ago

Gemma 4

Collection

8 items • Updated 14 days ago • 627

upvoted a collection 20 days ago

Nemotron-Post-Training-v3

Collection

Collection of datasets used in the post-training phase of Nemotron Nano and Super v3. • 28 items • Updated 2 days ago • 123

upvoted a collection 23 days ago

Open Pangram

Collection

Open models and datasets based on Pangram's ICLR 2026 EditLens paper licensed for noncommercial use ONLY under CC BY-NC-SA 4.0 • 4 items • Updated 23 days ago • 9

upvoted a collection 25 days ago

CodeScout

Collection

RL-trained code search agents (1.7B, 4B, 14B) that outperform 2–18× larger models using only a Unix terminal. 📄 arxiv.org/abs/2603.17829 • 12 items • Updated 29 days ago • 7

upvoted an article about 1 month ago

view article

Article

Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries

• 
• 
• +7

Mar 10

•

126

upvoted a collection about 1 month ago

Distil Efficiency Benchmarks

Collection

Collection of models used in the blog post www.distillabs.ai/blog/the-10x-inference-tax-you-dont-have-to-pay • 9 items • Updated Mar 2 • 3

upvoted 5 collections about 2 months ago

Quantized Qwen3.5

Collection

Verified models. Compatible with Transformers v5.3 and vLLM v0.16.1rc1 (nightly). Under evaluation. • 9 items • Updated Mar 12 • 9

REAM

Collection

Compressed MoE models with a reduced number of experts. See additional models at https://huggingface.co/bknyaz. • 11 items • Updated 7 days ago • 5

Cerebras REAP

Collection

Sparse MoE models compressed using REAP (Router-weighted Expert Activation Pruning) method • 30 items • Updated Feb 25 • 136

Qwen3.5

Collection

21 items • Updated Mar 9 • 1.53k

gliner2 family

Collection

GLiNER2 extends the original GLiNER architecture to support multi-task information extraction with a schema-driven interface. This base model provid • 4 items • Updated Feb 10 • 42