LLM EXPLORER 59,420 MODELS INDEXED

Cybertron V4 Qw7B MGS by fblgit

By fblgit · 10 downloads
HF Repo ↗ Copy URL

Cybertron V4 Qw7B MGS is an open-source language model by fblgit. Features: 7b LLM, VRAM: 15.2GB, Context: 128K, License: other, Instruction-Based, LLM Explorer Score: 0.24.

  Arxiv:2406.08464   Arxiv:2407.10671   Arxiv:2410.21228 Base model:finetune:qwen/qwen2... Base model:qwen/qwen2.5-7b-ins...   Conversational Dataset:magpie-align/magpie-qw...   Deploy:azure   En   Endpoints compatible   Generated from trainer   Instruct   Model-index   Qwen2   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/fblgit/cybertron-v4-qw7B-MGS 

Cybertron V4 Qw7B MGS Parameters and Internals

LLM NameCybertron V4 Qw7B MGS
Repository 🤗https://huggingface.co/fblgit/cybertron-v4-qw7B-MGS 
Base Model(s)  Qwen/Qwen2.5-7B-Instruct   Qwen/Qwen2.5-7B-Instruct
Model Size7b
Required VRAM15.2 GB
Updated2026-08-07
Maintainerfblgit
Model Typeqwen2
Instruction-BasedYes
Model Files  5.0 GB: 1-of-4   4.9 GB: 2-of-4   5.0 GB: 3-of-4   0.3 GB: 4-of-4
Supported Languagesen
Model ArchitectureQwen2ForCausalLM
Licenseother
Context Length131072
Model Max Length131072
Transformers Version4.45.2
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size152064
Torch Data Typebfloat16
Errorsreplace

Best Alternatives to Cybertron V4 Qw7B MGS

Best Alternatives
Context / RAM
Downloads
Likes
Qwen2.5 7B Instruct 1M986K / 15.4 GB52534373
Hush Qwen2.5 7B RP V1.4 1M986K / 15.2 GB102
Qwen2.5 7B RRP 1M986K / 15.2 GB56
Qwen2.5 7B CelestialHarmony 1M986K / 14.8 GB87
Q2.5 Instruct 1M Harmony986K / 15.2 GB42
Impish QWEN 7B 1M986K / 15.2 GB46
COCO 7B Instruct 1M986K / 15.2 GB94
Apollo 7B986K / 15.2 GB80
Graph R1 7B986K / 15.3 GB73
Qwen2.5 7B Sky R1 Mini986K / 15.2 GB60
Note: green Score (e.g. "73.2") means that the model is better than fblgit/cybertron-v4-qw7B-MGS.