Updated to most recent models and add ts in logging

docker-compose.yml

@@ -1,6 +1,6 @@
 services:
     jaycoach:
-        image: jaycoach
+        build: .
         environment:
             - HF_TOKEN=$HF_TOKEN
         ports:

server.js

@@ -2,15 +2,16 @@ const express 	= require('express')
 const app 		= express()
 const port 		= 8080
 
-process.on('uncaughtException', err => console.log('JayCoach:Exception:', err))
+process.on('uncaughtException', err => log('JayCoach:Exception:', err))
 
 const hfToken = process.env.HF_TOKEN
 
 
 const MODELS = {
 	
-	'phi3': {
-		name: 'microsoft/Phi-3-mini-128k-instruct'
+	
+	'phi3.5': {
+		name: 'microsoft/Phi-3.5-mini-instruct'
 		,prompt: function(prompt){
 			return [
 				"<|user|>"
@@ -21,22 +22,21 @@ const MODELS = {
 		}
 	}
 	
-	,'phi3mini4': {
-		name: 'microsoft/Phi-3-mini-4k-instruct'
+	
+	,'lama3.1': {
+		name: 'meta-llama/Meta-Llama-3.1-8B-Instruct'
 		,prompt: function(prompt){
 			return [
-				"<|user|>"
-				,prompt
-				+"<|end|>"
-				,"<|assistant|>"
+				"<|start_header_id|>user<|end_header_id|>"
+				,prompt+"<|eot_id|><|start_header_id|>assistant<|end_header_id|>"
+				+""
 			].join("\n")
 		}
 	}
 	
 	
-	
-	,'lama': {
-		name: 'meta-llama/Meta-Llama-3-8B-Instruct'
+	,'lama3.2v': {
+		name: 'meta-llama/Llama-3.2-11B-Vision-Instruct'
 		,prompt: function(prompt){
 			return [
 				"<|start_header_id|>user<|end_header_id|>"
@@ -46,8 +46,11 @@ const MODELS = {
 		}
 	}
 	
-	,'mistral':{
-		name: 'mistralai/Mistral-7B-Instruct-v0.3'
+	
+	
+	
+	,'mistral-small':{
+		name: 'mistralai/Mistral-Small-Instruct-2409'
 		,prompt: function(prompt){
 			return [
 				"[INST]"+prompt+"[/INST]"
@@ -56,6 +59,7 @@ const MODELS = {
 		
 	}
 	
+	
 	,'gemma':{
 		name: 'google/gemma-1.1-7b-it'
 		,prompt: function(prompt){
@@ -71,6 +75,13 @@ const MODELS = {
 	
 }
 
+function log(m){
+	let d = new Date();
+	let diso = d.toISOString();
+	
+	console.log(`${diso} ${m}`)
+}
+
 
 if(!hfToken){
 	throw new Error('NO TOKEN!');
@@ -177,6 +188,7 @@ async function GetModelAnswer(model, prompt){
 	while(i--){ // pra evitar um loop infinito, vai girar no maximo o numero de models...
 		let ModelConfig = MODELS[model];
 		let MyStats = ModelConfig.stats;
+		log("Stats:");
 		console.log(MyStats);
 		let InferenceApi = 'https://api-inference.huggingface.co/models/'  + ModelConfig.name;
 		
@@ -193,7 +205,7 @@ async function GetModelAnswer(model, prompt){
 			}
 		}
 		
-		console.log("Falando com a IA 🤖", model, ModelConfig.name)
+		log("Falando com a IA 🤖", model, ModelConfig.name)
 		MyStats.total++;
 		let StartTime = new Date();
 		const response = await fetch(
@@ -206,11 +218,11 @@ async function GetModelAnswer(model, prompt){
 		);
 		let EndTime = new Date();
 		let ElapsedTime = EndTime.getTime()  - StartTime.getTime();
-		console.log("Total", ElapsedTime);
+		log("Total", ElapsedTime);
 		
 		if(response.status != 200){
 			MyStats.erros++;
-			console.log('FAILED: Escolhendo outro...', response.status)
+			log('FAILED: Escolhendo outro...', response.status)
 			if(StartIndex == null)
 				StartIndex = ModelList.map(m => m.id).indexOf(model);
 			
@@ -220,12 +232,12 @@ async function GetModelAnswer(model, prompt){
 				NextIndex = 0;
 		
 			if(NextIndex == StartIndex){
-				console.log("Fiz de tudo, mas não deu bom :(");
+				log("Fiz de tudo, mas não deu bom :(");
 				throw new Error('SOME_SHIT_HAPPENS');
 			}
 			
 			model = ModelList[NextIndex].id;
-			console.log("Tentando com o ",model);
+			log("Tentando com o ",model);
 			continue;
 		}
 		
@@ -240,7 +252,7 @@ async function GetModelAnswer(model, prompt){
 			if(MyStats.erros < 0) MyStats.erros = 0;
 		}
 		
-		console.log("Ok, lendo o json...", response.status);
+		log("Ok, lendo o json...", response.status);
 		const result = await response.json();
 		LastWorkedModel = model;
 		return {
@@ -331,7 +343,7 @@ async function Prompt(opts){
 		---
 	`
 	
-	console.log("PromptLength: ", prompt.length, prompt);
+	log("PromptLength: ", prompt.length, prompt);
 	let answer = await GetModelAnswer(model, prompt);
 	return answer;
 }
@@ -368,7 +380,7 @@ app.get('/error', async (req, res) => {
 	
 	let ModelInfo = MODELS[result.model]
 	
-	console.log("Respondido", TotalMs, result);
+	log("Respondido", TotalMs, result);
 	
 	let resp = result.result;
 	if(!resp || !Array.isArray(resp)){
@@ -393,8 +405,9 @@ app.get('/error', async (req, res) => {
 		ModelInfo.stats.erros += 0.05;
 	}
 	
-	console.log("FullResp:"+gentext);
-	console.log("Final:"+txtFinal);
+
+	log("FullResp:"+gentext);
+	log(`Final:${txtFinal}`);
 	res.json({text:txtFinal, model:result.model, TotalMs})
 })
 
@@ -421,5 +434,5 @@ app.use(function(err, req, res, next) {
 
 
 app.listen(port, () => {
-  console.log(`JayCoach running`)
+  log(`JayCoach running`)
 })