AarambhDevHub/aarambh-studio
🦀 Decoder-only LLM built from scratch in pure Rust using Candle — no Python, no PyTorch. Gated DeltaNet + sparse attention, fine-grained MoE, native video/document understanding, long-horizon tool agents, quantization-aware training. Scales: Tiny (25M) to Large (1.3B).
ai-agentscandledecoder-onlydeep-learningdorafrom-scratchggufgrpolarge-language-modelllmloramachine-learningmixture-of-expertsmultimodal-llmquantizationquantization-aware-trainingrustself-learningsparse-attentionthinking-llm
Created
Jun 2026
Total stars
112
Last 7 days
0
Last 30 days
+2
Last 12 months
+112
README badge
[](https://rustystars.dev/AarambhDevHub/aarambh-studio)