Nemotron 3.5 Bolt E4B Artemis GGUF
The GGUF files in this repository are a modified version of NVIDIA Nemotron 3.5 Bolt E4B, developed by NVIDIA Corporation
(https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-BF16, revision 49843eed8db1f1c616c47af75cfd36e07aec0667).
NVIDIA post-trained that model from Gemma 4 E4B by Google (https://huggingface.co/google/gemma-4-E4B-it). Use of the
Gemma 4 model is governed by the Apache License, Version 2.0 (https://www.apache.org/licenses/LICENSE-2.0).
Changes: the text model was converted to GGUF and quantized for llama.cpp by Turing Intelligence Technology Limited
(TurinTech AI). The vision and audio encoders are included in the separate projector files mmproj-Q8_0.gguf and
mmproj-F16.gguf, converted from NVIDIA's weights with llama.cpp. MTP/dflash-Nemotron-3.5-Bolt-E4B-Q8_0.gguf is a Q8_0
quantization of NVIDIA's DFlash drafter for this model
(https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB-DFlash, revision
0949a2eb7152ec6fda1a91ff9302aecbcd464fbf, OpenMDW License Agreement version 1.1). This is an independent TurinTech
conversion, not affiliated with or endorsed by NVIDIA or Google.
This Ollama package contains the text model only; the vision and audio projectors and the DFlash drafter are on
Hugging Face (turintech/Nemotron-3.5-Bolt-E4B-Artemis-GGUF).
NVIDIA Nemotron 3.5 Bolt E4B and this modified version are made available under the OpenMDW License Agreement,
version 1.1 (https://openmdw.ai/license/1-1/), a copy of which is in LICENSE.
Quantization used importance data computed from public text: WikiText-2 raw train (Salesforce/wikitext,
wikitext-2-raw-v1; CC BY-SA 3.0 / GFDL), OpenAssistant/oasst1 train conversations (Apache-2.0) and CPython standard
library source (Python Software Foundation License). Importance matrices are provided only for the Q3_K_XL and Q4_K_XL
files, as Nemotron-3.5-Bolt-E4B-Artemis-Q3_K_XL-imatrix.gguf_file and
Nemotron-3.5-Bolt-E4B-Artemis-Q4_K_XL-imatrix.gguf_file. No calibration text is included in these files.
Please keep this NOTICE file when redistributing these files or derivatives.
Request (not a licence condition): if you serve these files to others, for example through an API, an inference provider
or OpenRouter, please mention TurinTech (https://turintech.ai) as the source of the quantization.