YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
v6e Best LLM - Trained on Trillium TPU
- Architecture: SmolLM2-360M (Llama, 960 hidden, 32 layers, 409M params)
- Dataset: FineWeb-Edu streaming
- Optimizer: AdamW + Warmup Cosine (peak 3e-4)
- Precision: bfloat16 on TPU v6e-1 (918 TFLOPS), 31GB HBM
- Budget: $39 at $3.05/h => 12.78h, target 2.29B tokens
- Hardware: AMD EPYC 9B14, 172GB RAM, TPU v6e-1
- Current step: 0 Trained by Muse Code.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support