LLM EXPLORER 57,252 MODELS INDEXED

Ising Calibration 1.5 31B BF16 by nvidia

By nvidia · 892 downloads
HF Repo ↗ Copy URL

Ising Calibration 1.5 31B BF16 is an open-source language model by nvidia. Features: 31b LLM, VRAM: 62.5GB, License: other.

  Arxiv:2604.25884   Bf16   Calibration   Conversational   Endpoints compatible   Gemma4   Image-text-to-text   Ising calibration   Nvidia   Quantum   Region:us   Safetensors   Sharded   Tensorflow   Vision   Vision-language
Model Card on HF 🤗: https://huggingface.co/nvidia/Ising-Calibration-1.5-31B-BF16 

Ising Calibration 1.5 31B BF16 Parameters and Internals

LLM NameIsing Calibration 1.5 31B BF16
Repository 🤗https://huggingface.co/nvidia/Ising-Calibration-1.5-31B-BF16 
Base Model(s)  google/gemma-4-31b   google/gemma-4-31b
Model Size31b
Required VRAM62.5 GB
Updated2026-08-10
Maintainernvidia
Model Typegemma4
Model Files  49.9 GB: 1-of-2   12.6 GB: 2-of-2
Model ArchitectureGemma4ForConditionalGeneration
Licenseother
Transformers Version5.6.0
Tokenizer ClassGemmaTokenizer
Padding Token<pad>

Best Alternatives to Ising Calibration 1.5 31B BF16

Best Alternatives
Context / RAM
Downloads
Likes
Gemma 4 31B It0K / 62.6 GB107066252780
Gemma 4 31B It0K / 62.6 GB3033922
Gemma 4 31B IT NVFP40K / 32.6 GB2773900547
Gemma 4 31B It Scotoma 20K / 62.5 GB22138
Gemma 4 31B It FP8 Block0K / 33.3 GB379858244
Gemma 4 31B It NVFP40K / 23.3 GB111553957
Gemma 4 31B It Abliterated0K / 62.5 GB75135524
Gemma 4 Dark Thoughts 31B0K / 62.6 GB19510
G4 MeroMero V2 31B0K / 62.5 GB19235
Gemma 4 31B It FP8 Dynamic0K / 33.3 GB22552223
Note: green Score (e.g. "73.2") means that the model is better than nvidia/Ising-Calibration-1.5-31B-BF16.