convaiinnovations/laya
Text Classification ā¢ 0.4B ā¢ Updated ā¢ 56.6k ā¢ ā¢ 5.82k
Afolabiwilson
Emeritus-21
AI & ML interests
None yet
Recent Activity
reacted to SeaWolf-AI's post with š„ 11 days ago
š We pretrained a language model from scratch on an inference-only NPU, with real backpropagation.
š¤ Model: https://huggingface.co/FINAL-Bench/VKORE-46M-NPU š Technical write-up: https://huggingface.co/blog/FINAL-Bench/backpropagation-on-an-inference-only-npu-pretraini
"NPUs are inference-only, you can't train on them." We built a counterexample: VKORE-46M-NPU, trained from scratch on a first-gen (2025) inference NPU, 32 chips, no pretrained weights. A 46M decoder-only model (RoPE Ā· SwiGLU Ā· RMSNorm Ā· GQA), char-level, TinyStories.
š Why trust it: a falling-loss curve proves nothing. So we trained the same model under identical init, data, and schedule on the NPU and on a GPU reference. Ā· Loss trajectories correlate at 0.99 Ā· Throughput converged to a comparable level The NPU backward pass and optimizer updates do numerically the same work as the GPU, not an approximation that merely trends down.
š§© One accelerator now covers pretraining Ā· LoRA Ā· inference Ā· quantization. We call this layer VKORE: "CUDA for NPUs."
āļø Public: weights, model card, NPU-vs-GPU comparison. Private (patent-pending, paper to follow): the core training technique. 46M char-level is toy scale on purpose: it proves the training path, not model quality. Larger models and real data are next.
Feedback from folks working on custom autograd backends and low-precision training kernels on non-GPU accelerators is very welcome. š liked a model about 1 month ago
convaiinnovations/laya liked a Space about 1 month ago
Emeritus-21/Hand-written-text-recognition-testingOrganizations
liked a
model about 1 month ago
liked a
Space about 1 month ago
liked a
dataset 3 months ago
liked a
model 9 months ago
liked a
model about 1 year ago
liked 2
datasets about 1 year ago
