🔧 Change separator token to </s> (#127)

Co-authored-by: OlivierDehaene <[email protected]>

.env

@@ -12,7 +12,7 @@ PUBLIC_MODEL_ID=OpenAssistant/oasst-sft-6-llama-30b-xor # used to link to model
 PUBLIC_DISABLE_INTRO_TILES=false
 PUBLIC_USER_MESSAGE_TOKEN=<|prompter|>
 PUBLIC_ASSISTANT_MESSAGE_TOKEN=<|assistant|>
-PUBLIC_SEP_TOKEN=<|endoftext|>
+PUBLIC_SEP_TOKEN=</s>
 PUBLIC_PREPROMPT="Below are a series of dialogues between various people and an AI assistant. The AI tries to be helpful, polite, honest, sophisticated, emotionally aware, and humble-but-knowledgeable. The assistant is happy to help with almost anything, and will do its best to understand exactly what is needed. It also tries to avoid giving false or misleading information, and it caveats when it isn't entirely sure about the right answer. That said, the assistant is practical and really does its best, and doesn't let caution get too much in the way of being useful."
 
 # [{"endpoint": "https://api-inference.huggingface.co/models/...", authorization: "Bearer hf_<token>", weight: 1}] to load balance

src/routes/conversation/[id]/+page.svelte

@@ -6,7 +6,7 @@
 	import { textGenerationStream } from "@huggingface/inference";
 	import { invalidate } from "$app/navigation";
 	import { base } from "$app/paths";
-	import { PUBLIC_MAX_INPUT_TOKENS } from "$env/static/public";
+	import { PUBLIC_MAX_INPUT_TOKENS, PUBLIC_SEP_TOKEN } from "$env/static/public";
 	import { shareConversation } from "$lib/shareConversation";
 	import { UrlDependency } from "$lib/types/UrlDependency";
 	import { error } from "$lib/stores/errors";
@@ -43,7 +43,7 @@
 					truncate: parseInt(PUBLIC_MAX_INPUT_TOKENS),
 					watermark: false,
 					max_new_tokens: 1024,
-					stop: ["<|endoftext|>"],
+					stop: [PUBLIC_SEP_TOKEN],
 					return_full_text: false,
 				},
 			},

src/routes/conversation/[id]/summarize/+server.ts

@@ -1,4 +1,4 @@
-import { PUBLIC_MAX_INPUT_TOKENS } from "$env/static/public";
+import { PUBLIC_MAX_INPUT_TOKENS, PUBLIC_SEP_TOKEN } from "$env/static/public";
 import { buildPrompt } from "$lib/buildPrompt";
 import { collections } from "$lib/server/database.js";
 import { modelEndpoint } from "$lib/server/modelEndpoint.js";
@@ -34,7 +34,7 @@ export async function POST({ params, locals, fetch }) {
 		watermark: false,
 		max_new_tokens: 1024,
 		truncate: parseInt(PUBLIC_MAX_INPUT_TOKENS),
-		stop: ["<|endoftext|>"],
+		stop: [PUBLIC_SEP_TOKEN],
 		return_full_text: false,
 	};