Use ungated version of mistral tokenizer (#718)

This commit is contained in:
Joshua Lochner
2024-04-20 16:01:17 +02:00
committed by GitHub
parent 992f643e2a
commit 1a9964fb09
3 changed files with 3 additions and 3 deletions
+1 -1
View File
@@ -2986,7 +2986,7 @@ export class PreTrainedTokenizer extends Callable {
* ```javascript
* import { AutoTokenizer } from "@xenova/transformers";
*
* const tokenizer = await AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1");
* const tokenizer = await AutoTokenizer.from_pretrained("Xenova/mistral-tokenizer-v1");
*
* const chat = [
* { "role": "user", "content": "Hello, how are you?" },
+1 -1
View File
@@ -229,7 +229,7 @@ TOKENIZERS_WITH_CHAT_TEMPLATES = {
'basic',
],
'mistralai/Mistral-7B-Instruct-v0.1': [
'Xenova/mistral-tokenizer-v1': [
'basic',
],
+1 -1
View File
@@ -334,7 +334,7 @@ describe('Extra decoding tests', () => {
describe('Chat templates', () => {
it('should generate a chat template', async () => {
const tokenizer = await AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.1");
const tokenizer = await AutoTokenizer.from_pretrained("Xenova/mistral-tokenizer-v1");
const chat = [
{ "role": "user", "content": "Hello, how are you?" },