Downloads · 30 days
26
8% of all-time downloads
Laguna-M.1-JANG_2L
OsaurusAI/Laguna-M.1-JANG_2L
Laguna-M.1-JANG_2L is a text generation model from OsaurusAI. Use it when you need the model to write or continue text. It is set up for mlx. The card lists the license as apache-2.0.
All-time downloads
318
Public
Parameters
226B
86.6 GB on disk
Likes
1
Public
Repo makeup
Click a slice to open those files.
.safetensors86.6 GB · 100%
Parameter types
How the weights are stored.
U32226B · 100%
Try a prompt
Base models
Model card
From the Hugging Face model README
Laguna-M.1 · JANG_2L
Official OsaurusAI affine-quantized build of poolside/Laguna-M.1 (Apache-2.0) — a 225B / 23B-active agentic-coding MoE. Quantized by Osaurus to the JANG_2L profile (~81 GB), runs on a single 128 GB Apple-Silicon machine via Osaurus / vMLX.
Profile (JANG_2L): 2-bit gate/up, 3-bit down, 6-bit shared/dense, 8-bit attn/embed-head — quality 2-bit (affine, group-size 64; per-module bits in config.json).
Runtime
Requires a Laguna-aware runtime (Osaurus / vMLX / jang_tools.laguna) — Laguna is a custom architecture (per-element attention gating, YaRN RoPE, sigmoid+bias MoE routing, shared expert) not in stock loaders. All-full attention → standard KV cache. For clean chat, stop on </assistant> / 〈|EOS|〉 and use repetition_penalty≈1.15.
Provenance
- Base: poolside/Laguna-M.1 © poolside — Apache-2.0
- Quantization: Osaurus · JANG affine · JANG_2L · [email protected]
- Task
- Text Generation
- Library
- mlx
- Type
- laguna
- License
- apache-2.0
- Created
- Jun 18, 2026
- Updated
- Jun 18, 2026