Steffen Röcker PRO

sroecker's activity

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

upvoted a collection about 14 hours ago

Fixed Chat Templates for Qwen 3.5 & 3.6

Collection

Rewritten Jinja templates fixing 5 bugs in official Qwen 3.5/3.6. Works in LM Studio, llama.cpp, MLX, vLLM. • 1 item • Updated 17 days ago • 3

upvoted a paper 1 day ago

Achieving Gold-Medal-Level Olympiad Reasoning via Simple and Unified Scaling

Paper • 2605.13301 • Published 4 days ago • 137

upvoted a collection 4 days ago

SpecDrift

Collection

Models released as a part of Attention-Drift Paper, trained for deployment on production • 2 items • Updated 7 days ago • 2

upvoted a collection 8 days ago

Gemma 4 Assistant GGUF

Collection

Gemma 4 MTP assistant drafters as GGUF (F16/Q8_0/Q5_K_M/Q4_K_M/Q4_K_S). Speculative-decoding heads for the atomic-llama-cpp-turboquant fork. • 4 items • Updated 10 days ago • 10

upvoted a collection 11 days ago

8GB VRAM Local LLMs - Practitioner Tested

Collection

practitioner benchmarks on RTX 4060 Ti 8GB. dense models + MoE expert offload sweeps. LFM2 52.2 > Qwen3.6 35.4 > Gemma 4 29.3 tok/s. • 17 items • Updated about 11 hours ago • 4

upvoted a collection 12 days ago

Granite 4.1 Language Models

Collection

Efficient language models for multilingual generation, coding, RAG, and AI assistant workflows. • 6 items • Updated 17 days ago • 50

upvoted an article 12 days ago

view article

Article

Granite 4.1 LLMs: How They’re Built

• 

ibm-granite

•

17 days ago

• 70

upvoted a collection 12 days ago

APEX Quants (GGUF)

Collection

MoE models quantized with the APEX Quantization technique ( https://github.com/mudler/apex-quant ) • 30 items • Updated about 5 hours ago • 94

upvoted a collection 14 days ago

1930 Coder

Collection

Fine-tuning the Talkie 13B 1930 model on agentic trajectories • 4 items • Updated 11 days ago • 4

upvoted a collection 16 days ago

Qwen3.6

Collection

4 items • Updated 25 days ago • 350

upvoted a collection 18 days ago

Laguna XS.2

Collection

Designed for agentic coding and long-horizon work on a local machine. Apache 2.0. • 5 items • Updated 9 days ago • 20

upvoted 2 collections 19 days ago

privacy-filter

Collection

OpenAI's privacy-filter fine0tuned models • 6 items • Updated 11 days ago • 10

talkie-13b

Collection

talkie-1930-13b is a vintage language model trained on pre-1931 English-language text. See https://github.com/talkie-lm/talkie to run talkie. • 3 items • Updated 26 days ago • 52

upvoted a collection 21 days ago

DeepSeek V4

Collection

8 items • Updated 22 days ago • 7

upvoted 2 collections 23 days ago

DeepSeek-V4

Collection

4 items • Updated 23 days ago • 643

DR-Venus

Collection

5 items • Updated 22 days ago • 17

upvoted a paper 25 days ago

Pushing the Limits of Large Language Model Quantization via the Linearity Theorem

Paper • 2411.17525 • Published Nov 26, 2024 • 6

upvoted a collection 25 days ago

HIGGS

Collection

Models prequantized with [HIGGS](https://arxiv.org/abs/2411.17525) zero-shot quantization. Requires the latest `transformers` to run. • 18 items • Updated Feb 18 • 15

upvoted an article 25 days ago

view article

Article

AI and the Future of Cybersecurity: Why Openness Matters

• 
• 
• +1

meg, yjernite, clem

•

26 days ago

• 38

upvoted a collection 26 days ago

Kimi K2.5

Collection

Moonshot's most powerful model • 3 items • Updated 26 days ago • 65