angelslim

SkillAI & models

Tencent AngelSlim — accessible, comprehensive, and efficient toolkit for large model compression. Quantization (FP8/INT4/NVFP4/1.25-bit), pruning, speculative decoding (Eagle3), and diffusion model compression.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the angelslim skill

What this skill tells your AI

The instructions your AI receives, as published by mkurman/zorai in skills/scientific-skills/angelslim/SKILL.md and read by ahel’s review.

Overview

AngelSlim integrates mainstream compression algorithms into a unified framework with one-click access. Supports FP8/INT8/INT4/NVFP4/1.25-bit quantization, pruning, Eagle3 speculative decoding, and diffusion model compression for LLMs, VLMs, and audio models.

Installation

uv pip install angelslim

Basic Quantization (PTQ)

import angelslim as slim

# FP8 static quantization
model = slim.quantize(model, dtype="fp8_static", qconfig="default")

# INT4 GPTQ
model = slim.quantize(model, dtype="int4_gptq", dataset="wikitext2")

Compression Strategies

MethodPrecisionBest For
FP8-Static/Dynamic8-bitGeneral LLM deployment
INT4 GPTQ/AWQ/GPTAQ4-bitMemory-constrained serving
NVFP44-bit (NVIDIA)Blackwell GPUs
Sherry1.25-bitExtreme compression
STQ1_01.25-bitOn-device deployment

Speculative Decoding (Eagle3)

# Train Eagle3 draft model
slim.eagle3.train(model, draft_model_config)

# Inference with Eagle3
output = model.generate_with_eagle3(input_ids, max_new_tokens=256)

References

Signals

GitHub stars
324
Forks
26
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
angelslim
Source
github.com/mkurman/zorai