package kaun
Flax-inspired neural network library for OCaml
Install
dune-project
Dependency
Authors
Maintainers
Sources
raven-1.0.0.alpha1.tbz
sha256=8e277ed56615d388bc69c4333e43d1acd112b5f2d5d352e2453aef223ff59867
sha512=369eda6df6b84b08f92c8957954d107058fb8d3d8374082e074b56f3a139351b3ae6e3a99f2d4a4a2930dd950fd609593467e502368a13ad6217b571382da28c
doc/kaun.huggingface/Kaun_huggingface/Registry/index.html
Module Kaun_huggingface.Registry
Source
Registry of known model architectures with their loading functions
Source
type ('params, 'a, 'dev) model_spec = {
architecture : string;
(*Architecture name (e.g., "GPT2", "BERT")
*)config_file : string;
(*Config filename (e.g., "config.json")
*)weight_files : string list;
(*Weight filenames to try (e.g.,
*)"model.safetensors", "pytorch_model.bin"
)load_config : Yojson.Safe.t -> 'params;
(*Parse config JSON into model parameters
*)build_params : dtype:(float, 'a) Rune.dtype -> 'params -> 'a Kaun.params;
(*Build parameter tree from config
*)
}
Register a model architecture
Get a registered model spec
sectionYPositions = computeSectionYPositions($el), 10)"
x-init="setTimeout(() => sectionYPositions = computeSectionYPositions($el), 10)"
>