diff --git a/main.js b/main.js
index 95ac664..8bc6351 100644
--- a/main.js
+++ b/main.js
@@ -8004,7 +8004,7 @@ var SemanticCacheService = class _SemanticCacheService {
   async clearCache() {
     if (!this.config.enabled || !this.client) return;
     try {
-      await this.client.reset();
+      await this.client.deleteCollection({ name: this.config.collectionName });
       Logger.info("Semantic cache cleared", "semantic-cache");
     } catch (error) {
       const errorMessage = error instanceof Error ? error.message : String(error);
@@ -9448,31 +9448,54 @@ var ContentVectorizer = class {
     this.fetchFn = fetchFn ?? ((url, init) => fetch(url, init));
   }
   /**
-   * Generates embeddings for a content chunk
+   * Generates embeddings for a content chunk with retry logic
    */
   async vectorize(chunk) {
-    try {
-      const prompt = this.createPrompt(chunk);
-      const response = await this.fetchFn(`${this.ollamaUrl}/api/embeddings`, {
-        method: "POST",
-        headers: { "Content-Type": "application/json" },
-        body: JSON.stringify({
-          model: this.model,
-          prompt
-        })
-      });
-      if (!response.ok) {
-        throw new Error(`Embedding failed with status ${response.status}`);
+    const prompt = this.createPrompt(chunk);
+    const maxRetries = 3;
+    const baseDelay = 1e3;
+    for (let attempt = 0; attempt < maxRetries; attempt++) {
+      try {
+        if (attempt > 0) {
+          const delay = baseDelay * Math.pow(2, attempt - 1);
+          Logger.info(
+            `Retrying embedding (attempt ${attempt + 1}/${maxRetries}) after ${delay}ms`,
+            "indexing-pipeline"
+          );
+          await new Promise((resolve) => setTimeout(resolve, delay));
+        }
+        const response = await this.fetchFn(`${this.ollamaUrl}/api/embeddings`, {
+          method: "POST",
+          headers: { "Content-Type": "application/json" },
+          body: JSON.stringify({
+            model: this.model,
+            prompt
+          })
+        });
+        if (!response.ok) {
+          throw new Error(`Embedding failed with status ${response.status}`);
+        }
+        const data = await response.json();
+        if (!this.isEmbeddingResponse(data)) {
+          throw new Error("Invalid embedding response");
+        }
+        return data.embedding;
+      } catch (error) {
+        const errorMessage = error instanceof Error ? error.message : String(error);
+        Logger.warn(
+          `Embedding attempt ${attempt + 1} failed: ${errorMessage}`,
+          "indexing-pipeline"
+        );
+        if (attempt === maxRetries - 1) {
+          Logger.warn(
+            `Failed to generate embedding after ${maxRetries} attempts`,
+            "indexing-pipeline"
+          );
+          return [];
+        }
       }
-      const data = await response.json();
-      if (!this.isEmbeddingResponse(data)) {
-        throw new Error("Invalid embedding response");
-      }
-      return data.embedding;
-    } catch (error) {
-      Logger.warn(`Failed to generate embedding: ${String(error)}`, "indexing-pipeline");
-      return [];
     }
+    return [];
   }
   isEmbeddingResponse(data) {
     return typeof data === "object" && data !== null && Array.isArray(data.embedding) && data.embedding.every((value) => typeof value === "number");
@@ -10010,8 +10033,8 @@ var OllamaPlugin = class extends import_obsidian4.Plugin {
     const files = this.app.vault.getMarkdownFiles();
     Logger.info(`Starting background vault indexing for ${files.length} files...`, "main");
     let indexed = 0;
-    const BATCH_SIZE = 5;
-    const DELAY_MS = 100;
+    const BATCH_SIZE = 2;
+    const DELAY_MS = 500;
     for (let i = 0; i < files.length; i += BATCH_SIZE) {
       if (signal.aborted) {
         Logger.info("Vault indexing cancelled.", "main");
diff --git a/src/indexing-pipeline/vectorization.ts b/src/indexing-pipeline/vectorization.ts
index 62612c0..fbbd2e1 100644
--- a/src/indexing-pipeline/vectorization.ts
+++ b/src/indexing-pipeline/vectorization.ts
@@ -23,36 +23,60 @@ export class ContentVectorizer {
   }
 
   /**
-   * Generates embeddings for a content chunk
+   * Generates embeddings for a content chunk with retry logic
    */
   async vectorize(chunk: ContentChunk): Promise<number[]> {
-    try {
-      const prompt = this.createPrompt(chunk);
+    const prompt = this.createPrompt(chunk);
+    const maxRetries = 3;
+    const baseDelay = 1000;
 
-      const response = await this.fetchFn(`${this.ollamaUrl}/api/embeddings`, {
-        method: 'POST',
-        headers: { 'Content-Type': 'application/json' },
-        body: JSON.stringify({
-          model: this.model,
-          prompt: prompt,
-        }),
-      });
+    for (let attempt = 0; attempt < maxRetries; attempt++) {
+      try {
+        if (attempt > 0) {
+          const delay = baseDelay * Math.pow(2, attempt - 1);
+          Logger.info(
+            `Retrying embedding (attempt ${attempt + 1}/${maxRetries}) after ${delay}ms`,
+            'indexing-pipeline'
+          );
+          await new Promise((resolve) => setTimeout(resolve, delay));
+        }
 
-      if (!response.ok) {
-        throw new Error(`Embedding failed with status ${response.status}`);
+        const response = await this.fetchFn(`${this.ollamaUrl}/api/embeddings`, {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({
+            model: this.model,
+            prompt: prompt,
+          }),
+        });
+
+        if (!response.ok) {
+          throw new Error(`Embedding failed with status ${response.status}`);
+        }
+
+        const data: unknown = await response.json();
+        if (!this.isEmbeddingResponse(data)) {
+          throw new Error('Invalid embedding response');
+        }
+
+        return data.embedding;
+      } catch (error) {
+        const errorMessage = error instanceof Error ? error.message : String(error);
+        Logger.warn(
+          `Embedding attempt ${attempt + 1} failed: ${errorMessage}`,
+          'indexing-pipeline'
+        );
+        if (attempt === maxRetries - 1) {
+          Logger.warn(
+            `Failed to generate embedding after ${maxRetries} attempts`,
+            'indexing-pipeline'
+          );
+          return [];
+        }
       }
-
-      const data: unknown = await response.json();
-      if (!this.isEmbeddingResponse(data)) {
-        throw new Error('Invalid embedding response');
-      }
-
-      return data.embedding;
-    } catch (error) {
-      // Return empty array on failure to maintain compatibility
-      Logger.warn(`Failed to generate embedding: ${String(error)}`, 'indexing-pipeline');
-      return [];
     }
+
+    return [];
   }
 
   private isEmbeddingResponse(data: unknown): data is { embedding: number[] } {
diff --git a/src/main.ts b/src/main.ts
index bd17e90..2550640 100755
--- a/src/main.ts
+++ b/src/main.ts
@@ -166,8 +166,8 @@ export default class OllamaPlugin extends Plugin {
     Logger.info(`Starting background vault indexing for ${files.length} files...`, 'main');
 
     let indexed = 0;
-    const BATCH_SIZE = 5;
-    const DELAY_MS = 100;
+    const BATCH_SIZE = 2;
+    const DELAY_MS = 500;
 
     for (let i = 0; i < files.length; i += BATCH_SIZE) {
       if (signal.aborted) {
@@ -191,7 +191,7 @@ export default class OllamaPlugin extends Plugin {
         })
       );
 
-      // Small delay between batches to avoid overloading Ollama
+      // Delay between batches to avoid overloading Ollama
       if (i + BATCH_SIZE < files.length) {
         await new Promise((resolve) => setTimeout(resolve, DELAY_MS));
       }
diff --git a/src/semantic-cache.ts b/src/semantic-cache.ts
index d0a03b5..4e6eb72 100644
--- a/src/semantic-cache.ts
+++ b/src/semantic-cache.ts
@@ -92,7 +92,7 @@ export class SemanticCacheService {
     if (!this.config.enabled || !this.client) return;
 
     try {
-      await this.client.reset();
+      await this.client.deleteCollection({ name: this.config.collectionName });
       Logger.info('Semantic cache cleared', 'semantic-cache');
     } catch (error) {
       const errorMessage = error instanceof Error ? error.message : String(error);
diff --git a/tests/semantic-cache.test.ts b/tests/semantic-cache.test.ts
index 5f3c4cf..232b7f3 100644
--- a/tests/semantic-cache.test.ts
+++ b/tests/semantic-cache.test.ts
@@ -155,13 +155,15 @@ describe('SemanticCacheService', () => {
 
       await disabledCacheService.clearCache();
 
-      expect(mockChromaClient.reset).not.toHaveBeenCalled();
+      expect(mockChromaClient.deleteCollection).not.toHaveBeenCalled();
     });
 
-    it('should clear the cache via the ChromaClient', async () => {
+    it('should clear the cache via deleteCollection', async () => {
       await cacheService.clearCache();
 
-      expect(mockChromaClient.reset).toHaveBeenCalled();
+      expect(mockChromaClient.deleteCollection).toHaveBeenCalledWith({
+        name: mockCacheConfig.collectionName,
+      });
     });
   });
 });