Skip to content
View Akicou's full-sized avatar
🌐
🌐
8 followers · 6 following
  • 08:58 (UTC +02:00)

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair ExtraordinaireAchievement: YOLO

Achievements

Achievement: Pull Sharkx2Achievement: QuickdrawAchievement: Pair ExtraordinaireAchievement: YOLO

Block or report Akicou

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Akicou/README.md

Isaac · Akicou

App Dev Apprentice · ML Tinkerer · Open Source

HuggingFace Nayhein GGUForge

I'm part of Nayhein AI, an open research org. Most of my work is around LLM compression, quantization pipelines, and the tooling that makes running large models on consumer hardware less painful.

Projects

Repo Description
GGUF Forge Web app that automates the full HuggingFace → GGUF pipeline. Download, convert, quantize, upload. Hosted free at gguforge.com (Hosted one Temporarily Offline).
REAM-MoE Generic REAM/REAP expert compression for MoE LLMs. Supports 15+ families — Qwen3, DeepSeek V3, Kimi K2, MiniMax M2, Mixtral and more.
rl-coding-agent RL loop that trains an LLM into a coding agent. Self-generating problems, sandboxed multi-language execution, zero human labels.
hf-local-hub Local HuggingFace Hub alternative written in Go.

Models & Datasets

29 models on HuggingFace — REAP/REAM compressed MoEs and GGUF quants for models like MiniMax M2, Qwen3 235B, and Solar Open 100B.

→ huggingface.co/Akicou

Python · Go · FastAPI · llama.cpp · PyTorch · Docker

Pinned Loading

  1. automaticConversion automaticConversion Public

    Python 2 1

  2. ream ream Public

    This repo implements a generic version of the REAM algorithm described in “REAM: Compressing Mixture-of-Experts LLMs” and makes it easy to apply to different MoE model families such as Qwen, gpt-os…

    Python 1 1

  3. rl-coding-agent rl-coding-agent Public

    Infinite RL loop that trains an open-source LLM into a SOTA coding agent — self-generating problems, multi-language sandboxed execution, zero human labels.

    Python

  4. hf-local-hub hf-local-hub Public

    Go 1