From cda1786eede5103d211832b2d400836ade1d54dc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Radam=C3=A9s=20Ajna?= Date: Fri, 22 Sep 2023 14:31:23 -0700 Subject: [PATCH] smaller t5 models quantized (#934) --- candle-wasm-examples/t5/utils.js | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/candle-wasm-examples/t5/utils.js b/candle-wasm-examples/t5/utils.js index e45e7d1b9e..851d1b76d4 100644 --- a/candle-wasm-examples/t5/utils.js +++ b/candle-wasm-examples/t5/utils.js @@ -67,7 +67,7 @@ export async function generateText( } export const MODELS = { t5_small_quantized: { - size: "102 MB", + size: "64.4 MB", base_url: "https://huggingface.co/lmz/candle-quantized-t5/resolve/main/", model: "model.gguf", tokenizer: "tokenizer.json", @@ -135,7 +135,7 @@ export const MODELS = { }, flan_t5_base_quantized: { - size: "360 MB", + size: "263 MB", base_url: "https://huggingface.co/lmz/candle-quantized-t5/resolve/main/", model: "model-flan-t5-base.gguf", tokenizer: "tokenizer.json",