Merge pull request #97 from pufflyai/96-add-gpt-4-turbo

add gpt-4-turbo

lerna.json

@@ -1,4 +1,4 @@
 {
   "$schema": "node_modules/lerna/schemas/lerna-schema.json",
-  "version": "0.22.1"
+  "version": "0.23.0"
 }

packages/@pufflig/ps-chains/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pufflig/ps-chains",
-  "version": "0.22.0",
+  "version": "0.23.0",
   "license": "MIT",
   "main": "./dist/ps-chains.umd.js",
   "module": "./dist/ps-chains.es.js",
@@ -16,7 +16,7 @@
     "test": "jest"
   },
   "devDependencies": {
-    "@pufflig/ps-types": "^0.22.0",
+    "@pufflig/ps-types": "^0.23.0",
     "@types/react-dom": "^18.2.7",
     "immer": "^10.0.2",
     "prop-types": "^15.8.1",

packages/@pufflig/ps-models/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@pufflig/ps-models",
   "private": false,
-  "version": "0.22.0",
+  "version": "0.23.0",
   "description": "Configuration of models used in Prompt Studio",
   "files": [
     "dist"
@@ -16,7 +16,7 @@
   "author": "Pufflig AB",
   "license": "MIT",
   "devDependencies": {
-    "@pufflig/ps-types": "^0.22.0",
+    "@pufflig/ps-types": "^0.23.0",
     "typescript": "^5.2.2",
     "vite": "^4.3.9",
     "vite-plugin-dts": "^2.3.0"

packages/@pufflig/ps-models/src/models/openai.ts

@@ -207,6 +207,71 @@ export const openai_completion: ModelDefinition = {
 };
 
 export const openai_chat: ModelDefinition = {
+  "gpt-4-1106-preview": {
+    modelId: "gpt-4-1106-preview",
+    description: "More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat.",
+    settings: openai_settings,
+    streaming: true,
+    contextLength: 4096,
+    parameters: [
+      {
+        id: "temperature",
+        type: "number",
+        name: "Temperature",
+        max: 2,
+        min: 0,
+        step: 0.1,
+        defaultValue: 0.4,
+        description:
+          "What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.",
+      },
+      {
+        id: "max_tokens",
+        type: "number",
+        name: "Max Tokens",
+        // although the documentation says the model should support 8192 tokens, it actually supports 4096
+        max: 4096,
+        min: 1,
+        step: 20,
+        defaultValue: 1024,
+        description:
+          "The maximum number of tokens to generate in the completion. The total length of input tokens and generated tokens is limited by the model's context length.",
+      },
+      {
+        id: "top_p",
+        type: "number",
+        name: "Top P",
+        max: 1,
+        min: 0,
+        step: 0.1,
+        defaultValue: 1,
+        description:
+          "An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.",
+      },
+      {
+        id: "frequency_penalty",
+        type: "number",
+        name: "Frequency penalty",
+        max: 2,
+        min: -2,
+        step: 0.1,
+        defaultValue: 0,
+        description:
+          "Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.",
+      },
+      {
+        id: "presence_penalty",
+        type: "number",
+        name: "Presence penalty",
+        max: 2,
+        min: -2,
+        step: 0.1,
+        defaultValue: 0,
+        description:
+          "Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.",
+      },
+    ],
+  },
   "gpt-4": {
     modelId: "gpt-4",
     description: "More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat.",

packages/@pufflig/ps-nodes-config/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@pufflig/ps-nodes-config",
   "private": false,
-  "version": "0.22.0",
+  "version": "0.23.0",
   "description": "Configuration files for nodes used in prompt studio.",
   "files": [
     "dist"
@@ -16,10 +16,10 @@
   "author": "Pufflig AB",
   "license": "MIT",
   "dependencies": {
-    "@pufflig/ps-models": "^0.22.0"
+    "@pufflig/ps-models": "^0.23.0"
   },
   "devDependencies": {
-    "@pufflig/ps-types": "^0.22.0",
+    "@pufflig/ps-types": "^0.23.0",
     "@types/jest": "^29.5.8",
     "jest": "^29.7.0",
     "ts-jest": "^29.1.0",

packages/@pufflig/ps-nodes-config/src/adapters/document_check/document_check.ts

@@ -1,11 +1,11 @@
-import { chat_models, completion_models, default_completion_model } from "@pufflig/ps-models";
 import { NodeConfig } from "@pufflig/ps-types";
+import { default_model, models } from "../../constants";
 
 export const documentCheckNodeType = "modifier/document_check" as const;
 
 export const documentCheck: NodeConfig = {
-  name: "Document Check",
-  description: "Run a checklist or extract information from a document.",
+  name: "Document Checklist",
+  description: "Run a checklist on a document.",
   tags: ["adapter", "document", "text"],
   status: "stable",
   execution: {
@@ -33,12 +33,12 @@ export const documentCheck: NodeConfig = {
   inputs: [
     {
       id: "model",
-      name: "Model",
+      name: "AI Settings",
       description: "The model to use",
       type: "model",
-      definition: { ...completion_models, ...chat_models },
+      definition: models,
       defaultValue: {
-        modelId: default_completion_model,
+        modelId: default_model,
         parameters: {},
       },
     },

packages/@pufflig/ps-nodes-config/src/adapters/llm/llm_completion.ts

@@ -1,5 +1,6 @@
-import { default_completion_model, completion_models } from "@pufflig/ps-models";
+import { models } from "@pufflig/ps-models";
 import { NodeConfig } from "@pufflig/ps-types";
+import { default_model } from "../../constants";
 
 export const llmCompletionNodeType = "adapter/llm_completion" as const;
 
@@ -34,12 +35,12 @@ export const llmCompletionConfig: NodeConfig = {
   inputs: [
     {
       id: "model",
-      name: "Model",
+      name: "AI Settings",
       description: "The model to use",
       type: "model",
-      definition: completion_models,
+      definition: models,
       defaultValue: {
-        modelId: default_completion_model,
+        modelId: default_model,
         parameters: {},
       },
     },

packages/@pufflig/ps-nodes-config/src/constants.ts

@@ -1 +1,17 @@
+import { chat_models, completion_models } from "@pufflig/ps-models";
+
 export const default_model = "gpt-3.5-turbo-instruct";
+
+export const available_models = [
+  "gpt-3.5-turbo-instruct",
+  "gpt-4-1106-preview",
+  "anthropic/claude-2",
+  "meta-llama/llama-2-13b-chat",
+];
+
+export const models = Object.values({ ...completion_models, ...chat_models })
+  .filter((model) => available_models.includes(model.modelId))
+  .reduce((acc, model) => {
+    acc[model.modelId] = model;
+    return acc;
+  }, {} as Record<string, any>);

packages/@pufflig/ps-nodes/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@pufflig/ps-nodes",
   "private": false,
-  "version": "0.22.1",
+  "version": "0.23.0",
   "description": "Collection of nodes used in Prompt Studio",
   "files": [
     "dist"
@@ -17,7 +17,7 @@
   "author": "Pufflig AB",
   "license": "MIT",
   "devDependencies": {
-    "@pufflig/ps-types": "^0.22.0",
+    "@pufflig/ps-types": "^0.23.0",
     "@types/jest": "^29.5.8",
     "@types/lodash": "^4.14.196",
     "@types/mustache": "^4.2.2",
@@ -33,9 +33,9 @@
   },
   "dependencies": {
     "@dqbd/tiktoken": "^1.0.7",
-    "@pufflig/ps-models": "^0.22.0",
-    "@pufflig/ps-nodes-config": "^0.22.0",
-    "@pufflig/ps-sdk": "^0.22.0",
+    "@pufflig/ps-models": "^0.23.0",
+    "@pufflig/ps-nodes-config": "^0.23.0",
+    "@pufflig/ps-sdk": "^0.23.0",
     "axios": "^1.6.2",
     "langchain": "^0.0.193",
     "lodash": "^4.17.21",