LLM EXPLORER 61,755 MODELS INDEXED

Jarvis Titan V15 MoE Decoupled by dhanesh-hf

By dhanesh-hf · 252 downloads
HF Repo ↗ Copy URL

Jarvis Titan V15 MoE Decoupled is an open-source language model by dhanesh-hf. Features: 14.8b LLM, VRAM: 29.6GB, Context: 32K, License: other, MoE, Merged, LLM Explorer Score: 0.3.

  Merged Model Base model:dhanesh-hf/jarvis-t... Base model:finetune:dhanesh-hf...   Conversational   Custom code   Deepseek-moe   Frontier-reasoning   Jarvis-titan   Jarvis titan moe   Long-context   Moe   Pallas-tpu   Region:us   Safetensors   Sharded   Tensorflow   Titans-neural-memory   Tri-brid-memory
Model Card on HF 🤗: https://huggingface.co/dhanesh-hf/Jarvis-Titan-V15-MoE-Decoupled 

Jarvis Titan V15 MoE Decoupled Parameters and Internals

LLM NameJarvis Titan V15 MoE Decoupled
Repository 🤗https://huggingface.co/dhanesh-hf/Jarvis-Titan-V15-MoE-Decoupled 
Base Model(s)  Jarvis Titan V14 MoE Merged   dhanesh-hf/Jarvis-Titan-V14-MoE-Merged
Merged ModelYes
Model Size14.8b
Required VRAM29.6 GB
Updated2026-09-20
Maintainerdhanesh-hf
Model Typejarvis_titan_moe
Model Files  4.7 GB: 1-of-7   4.6 GB: 2-of-7   4.6 GB: 3-of-7   4.8 GB: 4-of-7   4.5 GB: 5-of-7   4.6 GB: 6-of-7   1.8 GB: 7-of-7
Model ArchitectureJarvisTitanMoEForCausalLM
Licenseother
Context Length32768
Model Max Length32768
Transformers Version5.12.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size152064
Errorsreplace