Oriel — AI & technology worth noticing, every day.

nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB

AI model · https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB

First seen by Oriel on .

AI summary

NVIDIA-Nemotron-3.5-Bolt-E4B-4GB is a Q4_K_M quantized 4B-effective-parameter language model post-trained from Gemma-4-E4B-it for low-memory local assistants, with improved multi-turn tool use and instruction following.

  • The model has 8B total parameters, supports up to 64K-token input and output contexts, and is intended for voice and device assistants, interactive personas, and embedded interfaces.
  • NVIDIA trained it with multi-environment GRPO reinforcement learning across instruction following, multi-step tool use, multi-turn conversations, and structured output.
  • It supports configurable reasoning, runtime thinking budgets, conversational mode, and tool calling; restricted reasoning budgets can truncate thought processes and require task-specific evaluation.
  • The page provides llama.cpp deployment examples with an optional companion DFlash drafter for speculative decoding.
  • The model is available for commercial and non-commercial use under OpenMDW-1.1; its Gemma 4 base is governed by Apache License 2.0.

#language models #llama.cpp #speculative decoding #reinforcement learning

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Likes

Total likes reported by the source.

  • Hugging Face Models: 370, latest observation ; +261 over observed period since

Downloads (30d)

Rolling 30-day downloads. A change in this value is not new downloads during the observed period.

  • Hugging Face Models: 103,117, latest observation ; +97,211 over observed period since

Aliases

SourceExternal IDURL
Hugging Face Modelsnvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GBhttps://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Bolt-E4B-4GB Read discussion