Merge pull request #126 from StampyAI/qa-endpoint

Add page for randomized user testing

api/main.py

@@ -11,6 +11,8 @@ from stampy_chat.settings import Settings
 from stampy_chat.chat import run_query
 from stampy_chat.callbacks import stream_callback
 from stampy_chat.citations import get_top_k_blocks
+from stampy_chat.db.session import make_session
+from stampy_chat.db.models import Rating
 
 
 # ---------------------------------- web setup ---------------------------------
@@ -96,5 +98,23 @@ def human(id):
 
 # ------------------------------------------------------------------------------
 
+@app.route('/ratings', methods=['POST'])
+@cross_origin()
+def ratings():
+    session_id = request.json.get('sessionId')
+    settings = request.json.get('settings', {})
+    score = request.json.get('score')
+
+    if not session_id or score is None:
+        return Response('{"error": "missing params}', 400, mimetype='application/json')
+
+    with make_session() as s:
+        s.add(Rating(session_id=session_id, score=score, settings=json.dumps(settings)))
+        s.commit()
+
+    return jsonify({'status': 'ok'})
+
+
+
 if __name__ == '__main__':
     app.run(debug=True, port=FLASK_PORT)

api/migrations/versions/5813982e9665_ratings_table.py

@@ -0,0 +1,33 @@
+"""Ratings table
+
+Revision ID: 5813982e9665
+Revises: 78806d965229
+Create Date: 2023-11-06 17:31:47.814226
+
+"""
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import mysql
+from stampy_chat.db.models import UUID
+
+# revision identifiers, used by Alembic.
+revision = '5813982e9665'
+down_revision = '78806d965229'
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        'ratings',
+        sa.Column('id', sa.Integer(), nullable=False),
+        sa.Column('session_id', UUID(length=16), nullable=False),
+        sa.Column('score', sa.Integer(), nullable=False),
+        sa.Column('comment', mysql.LONGTEXT(), nullable=True),
+        sa.Column('settings', mysql.LONGTEXT(), nullable=False),
+        sa.Column('date_created', sa.DateTime(), nullable=False),
+        sa.PrimaryKeyConstraint('id')
+    )
+
+def downgrade() -> None:
+    op.drop_table('rating')

api/src/stampy_chat/db/models.py

@@ -90,3 +90,26 @@ class Interaction(Base):
 
     def __repr__(self) -> str:
         return f"Interaction(session={self.session_id!r}, no={self.interaction_no!r}, query={self.query!r}, response={self.response!r})"
+
+
+class Rating(Base):
+    __tablename__ = "ratings"
+
+    id: Mapped[int] = mapped_column("id", primary_key=True)
+
+    # The session_id is set once per session, so can be easily used to extract whole histories
+    session_id: Mapped[str] = mapped_column(UUID(), default=uuid.uuid4)
+
+    # the user provided score
+    score: Mapped[int] = mapped_column(Integer)
+
+    # An optional comment
+    comment: Mapped[Optional[str]] = mapped_column(LONGTEXT)
+
+    # The settings object, serialized to JSON
+    settings: Mapped[str] = mapped_column(LONGTEXT)
+
+    date_created: Mapped[datetime] = mapped_column(DateTime, default=func.now())
+
+    def __repr__(self) -> str:
+        return f"Rating(session={self.session_id!r}, score={self.score!r})"

web/src/components/settings.tsx

@@ -37,18 +37,18 @@ export const ChatSettings = ({
         onChange={(event: ChangeEvent) => {
           const value = (event.target as HTMLInputElement).value;
           const { maxNumTokens, topKBlocks } =
-            MODELS[value as keyof typeof MODELS];
+            MODELS[value as keyof typeof MODELS] || {};
           const prevNumTokens =
-            MODELS[settings.completions as keyof typeof MODELS].maxNumTokens;
+            MODELS[settings.completions as keyof typeof MODELS]?.maxNumTokens;
           const prevTopKBlocks =
-            MODELS[settings.completions as keyof typeof MODELS].topKBlocks;
+            MODELS[settings.completions as keyof typeof MODELS]?.topKBlocks;
 
           if (settings.maxNumTokens === prevNumTokens) {
             changeVal("maxNumTokens", maxNumTokens);
           } else {
             changeVal(
               "maxNumTokens",
-              Math.min(settings.maxNumTokens || 0, maxNumTokens)
+              Math.min(settings.maxNumTokens || 0, maxNumTokens || 0)
             );
           }
           if (settings.topKBlocks === prevTopKBlocks) {
@@ -86,7 +86,7 @@ export const ChatSettings = ({
         field="maxNumTokens"
         label="Tokens"
         min="1"
-        max={MODELS[settings.completions as keyof typeof MODELS].maxNumTokens}
+        max={MODELS[settings.completions as keyof typeof MODELS]?.maxNumTokens}
         updater={updateNum("maxNumTokens")}
       />
       <NumberInput

web/src/hooks/useSettings.ts

@@ -1,5 +1,5 @@
 import { useRouter } from "next/router";
-import { useState, useEffect } from "react";
+import { useState, useEffect, useCallback } from "react";
 
 import type { CurrentSearch, Mode, Entry, LLMSettings } from "../types";
 
@@ -42,7 +42,11 @@ const DEFAULT_PROMPTS = {
       "rather than just giving a formal definition.\n\n",
   },
 };
-export const MODELS = {
+interface Model {
+  maxNumTokens: number;
+  topKBlocks: number;
+}
+export const MODELS: { [key: string]: Model } = {
   "gpt-3.5-turbo": { maxNumTokens: 4095, topKBlocks: 10 },
   "gpt-3.5-turbo-16k": { maxNumTokens: 16385, topKBlocks: 30 },
   "gpt-4": { maxNumTokens: 8192, topKBlocks: 20 },
@@ -74,6 +78,13 @@ export const updateIn = (
   return obj;
 };
 
+const randomElement = (array: any[]) =>
+  array[Math.floor(Math.random() * array.length)];
+const randomFloat = (min: number, max: number) =>
+  Math.random() * (max - min) + min;
+const randomInt = (min: number, max: number) =>
+  Math.floor(randomFloat(min, max));
+
 /** Create a settings object in which all items in the `overrides` object will be parsed appropriately
  *
  *  `parsers` should be an object mapping settings fields to functions that will return a valid setting.
@@ -115,8 +126,8 @@ const SETTINGS_PARSERS = {
   mode: (v: string | undefined) => (v || "default") as Mode,
   completions: withDefault("gpt-3.5-turbo"),
   encoder: withDefault("cl100k_base"),
-  topKBlocks: withDefault(MODELS["gpt-3.5-turbo"].topKBlocks), //  the number of blocks to use as citations
-  maxNumTokens: withDefault(MODELS["gpt-3.5-turbo"].maxNumTokens),
+  topKBlocks: withDefault(MODELS["gpt-3.5-turbo"]?.topKBlocks), //  the number of blocks to use as citations
+  maxNumTokens: withDefault(MODELS["gpt-3.5-turbo"]?.maxNumTokens),
   tokensBuffer: withDefault(50), //  the number of tokens to leave as a buffer when calculating remaining tokens
   maxHistory: withDefault(10), //  the max number of previous items to use as history
   historyFraction: withDefault(0.25), //  the (approximate) fraction of num_tokens to use for history text before truncating
@@ -132,6 +143,27 @@ export const makeSettings = (overrides: LLMSettings) =>
     SETTINGS_PARSERS
   );
 
+const randomSettings = () => {
+  const completions = randomElement(Object.keys(MODELS));
+  const model = MODELS[completions] as Model;
+  const maxNumTokens = randomInt(
+    Math.floor(model.maxNumTokens * 0.3),
+    model.maxNumTokens
+  );
+  const historyFraction = randomFloat(0.2, 0.8);
+  const contextFraction = randomFloat(0.2, 0.9 - historyFraction);
+  return makeSettings({
+    completions,
+    maxNumTokens,
+    historyFraction,
+    contextFraction,
+    mode: randomElement(Object.keys(DEFAULT_PROMPTS.modes)) as Mode,
+    topKBlocks: randomInt(Math.floor(model.topKBlocks * 0.3), model.topKBlocks),
+    tokensBuffer: randomInt(10, 200),
+    maxHistory: randomInt(1, 20),
+  });
+};
+
 type ChatSettingsParams = {
   settings: LLMSettings;
   changeSetting: (path: string[], value: any) => void;
@@ -173,10 +205,13 @@ export default function useSettings() {
     setLoaded(router.isReady);
   }, [router]);
 
+  const randomize = useCallback(() => updateSettings(randomSettings()), []);
+
   return {
     settings,
     changeSetting,
     setMode,
     settingsLoaded,
+    randomize,
   };
 }

web/src/styles/globals.css

@@ -56,3 +56,12 @@ ol {
 .glossary-link {
   @apply underline hover:no-underline;
 }
+
+.rate-container {
+  margin-top: 20px;
+}
+
+.rate-button {
+  width: 2.5em;
+  margin: 0.5em;
+}