LLM EXPLORER 55,709 MODELS INDEXED

GPT Oss 20.3B Specialized Harmful Pruned MoE Only 31 Experts by AmanPriyanshu

By AmanPriyanshu · 8 downloads
HF Repo ↗ Copy URL

GPT Oss 20.3B Specialized Harmful Pruned MoE Only 31 Experts is an open-source language model by AmanPriyanshu. Features: 20.3b LLM, VRAM: 40.7GB, Context: 128K, License: apache-2.0, MoE, LLM Explorer Score: 0.19.

  Conversational Dataset:amanpriyanshu/gpt-oss-...   Domain-specific   Efficient   En   Expert-pruning   Gpt-oss   Gpt oss   Harmful   Mixture-of-experts   Moe   Openai   Pruned-model   Pytorch   Reasoning   Region:us   Safetensors   Sharded   Specialized   Tensorflow
Model Card on HF 🤗: https://huggingface.co/AmanPriyanshu/gpt-oss-20.3b-specialized-harmful-pruned-moe-only-31-experts 

GPT Oss 20.3B Specialized Harmful Pruned MoE Only 31 Experts Parameters and Internals

LLM NameGPT Oss 20.3B Specialized Harmful Pruned MoE Only 31 Experts
Repository 🤗https://huggingface.co/AmanPriyanshu/gpt-oss-20.3b-specialized-harmful-pruned-moe-only-31-experts 
Model Size20.3b
Required VRAM40.7 GB
Updated2026-07-17
MaintainerAmanPriyanshu
Model Typegpt_oss
Model Files  1.2 GB: 1-of-26   1.6 GB: 2-of-26   1.6 GB: 3-of-26   1.6 GB: 4-of-26   1.6 GB: 5-of-26   1.6 GB: 6-of-26   1.6 GB: 7-of-26   1.6 GB: 8-of-26   1.6 GB: 9-of-26   1.6 GB: 10-of-26   1.6 GB: 11-of-26   1.6 GB: 12-of-26   1.6 GB: 13-of-26   1.6 GB: 14-of-26   1.6 GB: 15-of-26   1.6 GB: 16-of-26   1.6 GB: 17-of-26   1.6 GB: 18-of-26   1.6 GB: 19-of-26   1.6 GB: 20-of-26   1.6 GB: 21-of-26   1.6 GB: 22-of-26   1.6 GB: 23-of-26   1.6 GB: 24-of-26   1.5 GB: 25-of-26   1.2 GB: 26-of-26
Supported Languagesen
Model ArchitectureGptOssForCausalLM
Licenseapache-2.0
Context Length131072
Model Max Length131072
Transformers Version4.55.0
Padding Token<|endoftext|>
Vocabulary Size201088