From 1f4a65675564b034f2d9c349424840cac3db03cf Mon Sep 17 00:00:00 2001
From: Sean Sundberg <seansund@us.ibm.com>
Date: Thu, 26 Oct 2023 11:30:26 -0500
Subject: [PATCH 1/2] Reduce find relevant passage concurrent requests to 2

Signed-off-by: Sean Sundberg <seansund@us.ibm.com>
---
 src/services/data-extraction/data-extraction.impl.ts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/services/data-extraction/data-extraction.impl.ts b/src/services/data-extraction/data-extraction.impl.ts
index 35e795d..b19ab04 100644
--- a/src/services/data-extraction/data-extraction.impl.ts
+++ b/src/services/data-extraction/data-extraction.impl.ts
@@ -14,7 +14,7 @@ import {first, GenAiModel, GenerativeResponse} from "../../utils";
 import pQueue from '../../utils/p-queue'
 import PQueue from "../../utils/p-queue";
 
-const queue = new PQueue({concurrency: 4});
+const queue = new PQueue({concurrency: 2});
 
 export interface DataExtractionBackendConfig {
     identityUrl: string;

From c2c95ed912132205b219c7730b8f4e69d40d8cfb Mon Sep 17 00:00:00 2001
From: Sean Sundberg <seansund@us.ibm.com>
Date: Thu, 26 Oct 2023 11:09:30 -0500
Subject: [PATCH 2/2] Update llm error handling for data extract

Signed-off-by: Sean Sundberg <seansund@us.ibm.com>
---
 .../data-extraction/data-extraction.impl.ts   | 25 +++++++++++++++----
 src/utils/gen-ai-model.ts                     |  2 +-
 2 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/src/services/data-extraction/data-extraction.impl.ts b/src/services/data-extraction/data-extraction.impl.ts
index b19ab04..83d10e7 100644
--- a/src/services/data-extraction/data-extraction.impl.ts
+++ b/src/services/data-extraction/data-extraction.impl.ts
@@ -219,11 +219,26 @@ export class DataExtractionImpl extends DataExtractionCsv<WatsonBackends, Contex
         const input = prompt + '\n\n' + text;
 
         const modelId = config.model || this.backendConfig.modelId;
-        const result: GenerativeResponse = await backends.wml.generate({
-            input,
-            modelId,
-            parameters,
-        });
+        const result: GenerativeResponse = await backends.wml
+            .generate({
+                input,
+                modelId,
+                parameters,
+            })
+            .then(result => {
+                if (result?.generatedText?.trim()) {
+                    return result
+                }
+
+                const fallbackModelId: string = 'meta-llama/llama-2-70b-chat'
+                console.log(`*** No information returned from generate. Trying again with ${fallbackModelId} model`)
+
+                return backends.wml.generate({
+                    input,
+                    modelId: fallbackModelId,
+                    parameters,
+                })
+            });
 
         console.log('2. Text generated from watsonx.ai:', {prompt, modelId, max_new_tokens, generatedText: result.generatedText, input})
 
diff --git a/src/utils/gen-ai-model.ts b/src/utils/gen-ai-model.ts
index 9608ad9..315bf3a 100644
--- a/src/utils/gen-ai-model.ts
+++ b/src/utils/gen-ai-model.ts
@@ -91,7 +91,7 @@ export class GenAiModel {
                 }
 
                 console.log('Error generating text: ', err);
-                throw err;
+                return {generatedText: '[Error]'}
             })
     }
 }
\ No newline at end of file