Add removing of duplicates from pipeline. Add to sort step. Move score logic to robertaMetrics node.

agent/nodes/robertaMetrics.ts

@@ -8,7 +8,12 @@ export const robertaMetrics: GraphNode<typeof MessagesState> = async (state) =>
   
   const result = await evaluateWithRoberta({answer})
   
+  let score = 0;
+  if (result.validProb > result.invalidProb) {
+    score = 0.7 + ((result.validProb - result.invalidProb)*0.3);
+  }
+  
   return {
-    messages: [ new AIMessage("ROBERTA:" + result)]
+    messages: [ new AIMessage("ROBERTA:" + score)]
   };
 };

agent/nodes/sort.ts

@@ -1,10 +1,14 @@
 import { GraphNode } from "@langchain/langgraph";
 import { MessagesState } from "../state";
 import { AIMessage } from "@langchain/core/messages";
+import { removeDuplicates } from "../tools/removeDuplicates";
 export const sort: GraphNode<typeof MessagesState> = async (state) => {
-  //not sure which will be better from API, just do both
-
   let current = state.proposedTriggerEvent;
+
+  // remove duplicates
+  await removeDuplicates(current)
+
+  // not sure which will be better from API, just do both
   current.sort((a, b) => ((b.score as number) ?? 0) - ((a.score as number) ?? 0));
 
 

agent/tools/removeDuplicates.ts

@@ -0,0 +1,44 @@
+import { pipeline, cos_sim } from "@huggingface/transformers";
+
+let featureExtractor = await pipeline(
+    "feature-extraction",
+    "Xenova/all-MiniLM-L6-v2"
+);
+
+export async function removeDuplicates(state: any) {
+    const embeddings: number[][] = [];
+
+    const outputs = await featureExtractor(
+        state.map(s => s.Event),
+        { pooling: "mean", normalize: true }
+    );
+
+    for (const o of outputs) {
+      embeddings.push(Array.from(o.data));
+    }
+
+    const len = state.length;
+    for (let i = 0; i < len; i++) {
+        for (let j = i + 1; j < len; j++) {
+            if (state[i].score === -1 || state[j].score === -1) continue;
+
+            const sim = cos_sim(embeddings[i], embeddings[j]);
+            console.log(sim)
+            if (sim > 0.55) {
+                const scoreI = state[i].score ?? 0;
+                const scoreJ = state[j].score ?? 0;
+
+                if (scoreI > scoreJ) {
+                    state[j].score = -1;
+                } else if (scoreJ > scoreI) {
+                    state[i].score = -1;
+                } else {
+                    // if equal, keep earlier
+                    state[j].score = -1;
+                }
+            }
+        }
+    }
+
+    return state;
+}

agent/tools/robertaCall.ts

@@ -4,15 +4,15 @@ export async function evaluateWithRoberta({
   answer
 }: {
   answer: string;
-}) {
+}): Promise<{ validProb: number; invalidProb: number; }> {
   const res = await axios.post("http://localhost:8000/evaluate", {
     answer
   });
   // console.log(res.data)
   const validProb = res.data["probabilities"][0][0]
-  const invalidProv = res.data["probabilities"][0][1]
+  const invalidProb = res.data["probabilities"][0][1]
 
-  return validProb > invalidProv ? 1 : 0;
+  return {validProb, invalidProb};
 }
 
 // let res = await evaluateWithRoberta({answer: "High-profile political downplaying of COVID-19 (examples: President Trump saying 'it will go away' in March–August 2020)"});