Qovaryx 50M Scratch Base is an open-source language model by tjarvis91. Features: 50m LLM, VRAM: 0.1GB, Context: 1K, License: apache-2.0, LLM Explorer Score: 0.31.
16gb-vram Bitnet Blackwell Causal-language-model Compact-ai Compact-llm Consumer-gpu Custom code Decoder Edge-ai En Flash-attention-2 From-scratch Gqa Grouped-query-attention Local-ai Low-rank Mixture-of-experts Moe Mtp Multi-token-prediction Nanogpt On-device Own-base Pretraining-template Pytorch Qovaryx Random-init Region:us Rtx-4090 Rtx-5070-ti Rtx-5090 Scratch-base Sm120 Small-language-model Sovereign-ai Sparse-moe Swiglu Ternary-swiglu Tiny-llm
Model Card on HF 🤗: https://huggingface.co/tjarvis91/qovaryx-50m-scratch-base
Qovaryx 50M Scratch Base Benchmarks
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
What is the LLM Explorer Rank (Score)
Qovaryx 50M Scratch Base Parameters and Internals
| LLM Name | Qovaryx 50M Scratch Base |
| Repository 🤗 | https://huggingface.co/tjarvis91/qovaryx-50m-scratch-base |
| Model Size | 50m |
| Required VRAM | 0.1 GB |
| Updated | 2026-06-04 |
| Maintainer | tjarvis91 |
| Model Type | qovaryx |
| Model Files | |
| Supported Languages | en |
| Model Architecture | QovaryxForCausalLM |
| License | apache-2.0 |
| Context Length | 1024 |
| Model Max Length | 1024 |
| Transformers Version | 5.2.0 |
| Tokenizer Class | PreTrainedTokenizerFast |
| Padding Token | <pad> |
| Vocabulary Size | 20242 |
| Torch Data Type | bfloat16 |
Rank the Qovaryx 50M Scratch Base Capabilities
🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟
What open-source LLMs or SLMs are you in search of? 54290 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a