hizrianraz/Laguna-S-2.1-GGUF overview
Laguna S 2.1 GGUF Personal GGUF mirror of poolside/Laguna S 2.1 https://huggingface.co/poolside/Laguna S 2.1 for Spark / llama.cpp paths. Not a re train. Not a…
Runs locally from ~89.44 GB disk (32 GB+ VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| laguna-s-2.1-Q4_K_M.gguf | GGUF | Q4_K_M | 89.44 GB | Download |
Model Details
| Model ID | hizrianraz/Laguna-S-2.1-GGUF |
|---|---|
| Author | hizrianraz |
| Pipeline | text-generation |
| License | other |
| Base model | poolside/Laguna-S-2.1 |
| Last modified | 2026-08-03T13:51:24.000Z |
Model README
---
license: other
license_name: openmdw-1.1
license_link: https://huggingface.co/poolside/Laguna-S-2.1/blob/main/LICENSE.md
language:
- en
library_name: gguf
pipeline_tag: text-generation
base_model: poolside/Laguna-S-2.1
base_model_relation: quantized
quantized_by: hizrianraz
tags:
- laguna
- laguna-s
- laguna-s-2.1
- poolside
- moe
- gguf
- quantized
- q4_k_m
- agentic-coding
- coding
- code
- code-generation
- tool-use
- function-calling
- agents
- llama.cpp
- dgx-spark
- nvidia
- cuda
- measured
- conversational
- text-generation
---
Laguna-S-2.1-GGUF
Personal GGUF mirror of poolside/Laguna-S-2.1 for Spark / llama.cpp paths.
> Not a re-train. Not a new architecture. Quantized-by hizrianraz from the upstream BF16 open weight.
> Prefer official poolside/Laguna-S-2.1-GGUF when it covers your need; this mirror exists for digests pinned by the Spark-Agentic pack.
Model parameters (upstream)
| | |
|--|--|
| Total | ~118B (HF count 117.56B BF16) |
| Active / token | ~8B |
| Experts | 256 routed (top-10) + 1 shared |
| Arch | MoE · LagunaForCausalLM · agentic coding / long-horizon |
| Source | poolside/Laguna-S-2.1 |
| Official GGUF | poolside/Laguna-S-2.1-GGUF |
| This repo | byte-identical official Q4_K_M mirror · not a re-quant · base_model_relation: quantized |
> Native Model tree: base_model → poolside/Laguna-S-2.1 · base_model_relation: quantized.
Files
laguna-s-2.1-Q4_K_M.gguf— Q4_K_M · pin in sibling packRELEASE/SHA256SUMS- See
SHA256SUMSand pack receipts for the authoritative digest
Model tree
- Base: poolside/Laguna-S-2.1
- Relation: quantized (GGUF)
- Deployment pack (runtime / measured path): hizrianraz/Laguna-S-2.1-Spark-Agentic
Model parameters (upstream)
From Poolside Laguna S 2.1:
- 118B total · ~8B activated per token
- 256 routed experts (top-10) + 1 shared
- Agentic coding / long-horizon design point
- Official GGUF also published under poolside/Laguna-S-2.1-GGUF
Honesty
diy_gguf=falsefor public OS path — this mirror pins published Q4; not a custom DIY invent- Smoke ≠ headline · verifier ≠ gate clearance
- Personal mirror · not Poolside · not Ainfera product marketing
- License: OpenMDW 1.1 — see upstream
LICENSE.md
License
OpenMDW-1.1 — upstream LICENSE. See NOTICE.
Run hizrianraz/Laguna-S-2.1-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models