LLM EXPLORER 62,696 MODELS INDEXED

Atomight V2.1 0.5B Inference by NovatasticRoScript

By NovatasticRoScript · 670 downloads
HF Repo ↗ Copy URL

Atomight V2.1 0.5B Inference is an open-source language model by NovatasticRoScript. Features: 0.5b LLM, VRAM: 1GB, Context: 32K, License: mit, LLM Explorer Score: 0.26.

  Arxiv:1803.05457   Arxiv:2110.14168   Arxiv:2402.03300   Arxiv:2412.15115   Atomightv2.1   Conversational   Dataset:allenai/ai2 arc Dataset:bespokelabs/bespoke-st...   Dataset:cais/mmlu   Dataset:openai/gsm8k Dataset:openai/openai humaneva...   En   Grpo   Lm-evaluation-harness   Mini-llm   Model-index   Qwen2   Reasoning   Region:us   Safetensors
Model Card on HF 🤗: https://huggingface.co/NovatasticRoScript/Atomight-V2.1-0.5B-Inference 

Atomight V2.1 0.5B Inference Parameters and Internals

LLM NameAtomight V2.1 0.5B Inference
Repository 🤗https://huggingface.co/NovatasticRoScript/Atomight-V2.1-0.5B-Inference 
Base Model(s)  Qwen/Qwen2.5-0.5B   Qwen/Qwen2.5-0.5B
Model Size0.5b
Required VRAM1 GB
Updated2026-07-26
MaintainerNovatasticRoScript
Model Typeqwen2
Model Files  1.0 GB
Supported Languagesen
Model ArchitectureQwen2ForCausalLM
Licensemit
Context Length32768
Model Max Length32768
Tokenizer ClassQwen2Tokenizer
Padding Token<|PAD_TOKEN|>
Vocabulary Size151936
Torch Data Typefloat16
Errorsreplace

Best Alternatives to Atomight V2.1 0.5B Inference

Best Alternatives
Context / RAM
Downloads
Likes
Reader Lm 0.5B250K / 1 GB188150
Qwen2 0.5B128K / 1 GB1216780169
Qwen2 0.5B Finetune Exp 1128K / 1 GB150
...0.5B FT EnhancedMUSREnsembleV3128K / 1 GB321
....5B FT V4 MUSR ENSEMBLE Mathis128K / 2 GB301
...0.5B FT MUSR ENSEMBLE V2Mathis128K / 1 GB250
...PRYMMAL 0.5B FT V4 MUSR Mathis128K / 1 GB621
Qwen2 0.5B Stf Imdb128K / 1 GB70
ECE PRYMMAL 0.5B FT V3 MUSR128K / 2 GB170
ECE PRYMMAL 0.5B FT V4 MUSR128K / 2 GB100
Note: green Score (e.g. "73.2") means that the model is better than NovatasticRoScript/Atomight-V2.1-0.5B-Inference.