Tivaphraen/LFM2-8B-A1B-smoltalk-GGUF overview
Ce dépôt contient les quantifications au format GGUF du modèle LiquidAI/LFM2 8B A1B smoltalk LoRA https://huggingface.co/LiquidAI/LFM2 8B A1B smoltalk LoRA . ⚙…
Runs locally from ~3.72 GB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| LFM2-8B-A1B-smoltalk-MXFP4_MOE.gguf | GGUF | GGUF | 4.42 GB | Download |
| LFM2-8B-A1B-smoltalk-Q3_K_M.gguf | GGUF | Q3_K_M | 3.72 GB | Download |
| LFM2-8B-A1B-smoltalk-Q4_K_M.gguf | GGUF | Q4_K_M | 4.70 GB | Download |
| LFM2-8B-A1B-smoltalk-Q5_K_M.gguf | GGUF | Q5_K_M | 5.51 GB | Download |
| LFM2-8B-A1B-smoltalk-Q6_K.gguf | GGUF | Q6_K | 6.38 GB | Download |
| LFM2-8B-A1B-smoltalk-Q8_0.gguf | GGUF | Q8_0 | 8.26 GB | Download |
Model Details
| Model ID | Tivaphraen/LFM2-8B-A1B-smoltalk-GGUF |
|---|---|
| Author | Tivaphraen |
| Pipeline | text-generation |
| License | other |
| Base model | LiquidAI/LFM2-8B-A1B-smoltalk-LoRA |
| Last modified | 2026-06-11T04:52:14.000Z |
Model README
---
license: other
license_name: lfm1.0
license_link: LICENSE
base_model:
- LiquidAI/LFM2-8B-A1B-smoltalk-LoRA
base_model_relation: quantized
tags:
- gguf
- llama.cpp
- lfm2
- lfm2-moe
- moe
- mixture-of-experts
- liquid-ai
- smoltalk
- lora-merged
datasets: HuggingFaceTB/smoltalk
language:
- en
- ar
- zh
- fr
- de
- ja
- ko
- es
pipeline_tag: text-generation
---
Ce dépôt contient les quantifications au format GGUF du modèle LiquidAI/LFM2-8B-A1B-smoltalk-LoRA.
⚙️ Paramètres de génération recommandés (LFM2)
Temperature = 0.3
Min_p = 0.15
Repetition_penalty = 1.05
🛠️ Méthode de quantization
merge_and_unload() de l'adaptateur LoRA dans le base model (PEFT >= 0.19.2.dev0, Transformers main)
Conversion HF → GGUF BF16 via convert_hf_to_gguf.py
Quantization via llama-quantize (llama.cpp)
📜 Licence
Hérite de la licence du base model : LFM Open License v1.0.
🙏 Remerciements
Liquid AI pour le base model et l'adaptateur LoRA
HuggingFaceTB pour le dataset SmolTalk
llama.cpp pour la quantization
Run Tivaphraen/LFM2-8B-A1B-smoltalk-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models