♻️ refactor: temperature range from 0 to 2 (lobehub#3355)

* feat: Update temperature settings for agent configuration * fix * All completed * temperature / 2 * fix * fix test * add test * ✅ test: fix test * 🚨 ci: fix lint --------- Co-authored-by: arvinxx <[email protected]>
trocker · Sep 8, 2024 · 4a9114e · 4a9114e
1 parent 988d744
commit 4a9114e
Show file tree

Hide file tree

Showing 27 changed files with 290 additions and 27 deletions.
diff --git a/src/const/settings/agent.ts b/src/const/settings/agent.ts
@@ -26,7 +26,7 @@ export const DEFAULT_AGENT_CONFIG: LobeAgentConfig = {
   params: {
     frequency_penalty: 0,
     presence_penalty: 0,
-    temperature: 0.6,
+    temperature: 1,
     top_p: 1,
   },
   plugins: [],

diff --git a/src/database/client/schemas/session.ts b/src/database/client/schemas/session.ts
@@ -32,7 +32,7 @@ export const AgentSchema = z.object({
     frequency_penalty: z.number().default(0).optional(),
     max_tokens: z.number().optional(),
     presence_penalty: z.number().default(0).optional(),
-    temperature: z.number().default(0.6).optional(),
+    temperature: z.number().default(1).optional(),
     top_p: z.number().default(1).optional(),
   }),
   plugins: z.array(z.string()).optional(),

diff --git a/src/features/AgentSetting/AgentModal/index.tsx b/src/features/AgentSetting/AgentModal/index.tsx
@@ -36,7 +36,7 @@ const AgentModal = memo(() => {
         tag: 'model',
       },
       {
-        children: <SliderWithInput max={1} min={0} step={0.1} />,
+        children: <SliderWithInput max={2} min={0} step={0.1} />,
         desc: t('settingModel.temperature.desc'),
         label: t('settingModel.temperature.title'),
         name: ['params', 'temperature'],

diff --git a/src/features/ChatInput/ActionBar/Temperature.tsx b/src/features/ChatInput/ActionBar/Temperature.tsx
@@ -24,7 +24,7 @@ const Temperature = memo(() => {
       content={
         <SliderWithInput
           controls={false}
-          max={1}
+          max={2}
           min={0}
           onChange={(v) => {
             updateAgentConfig({ params: { temperature: v } });

diff --git a/src/libs/agent-runtime/anthropic/index.test.ts b/src/libs/agent-runtime/anthropic/index.test.ts
@@ -155,7 +155,7 @@ describe('LobeAnthropicAI', () => {
           messages: [{ content: 'Hello', role: 'user' }],
           model: 'claude-3-haiku-20240307',
           stream: true,
-          temperature: 0.5,
+          temperature: 0.25,
           top_p: 1,
         },
         {},
@@ -192,7 +192,7 @@ describe('LobeAnthropicAI', () => {
           messages: [{ content: 'Hello', role: 'user' }],
           model: 'claude-3-haiku-20240307',
           stream: true,
-          temperature: 0.5,
+          temperature: 0.25,
           top_p: 1,
         },
         {},

diff --git a/src/libs/agent-runtime/anthropic/index.ts b/src/libs/agent-runtime/anthropic/index.ts
@@ -96,7 +96,10 @@ export class LobeAnthropicAI implements LobeRuntimeAI {
       messages: buildAnthropicMessages(user_messages),
       model,
       system: system_message?.content as string,
-      temperature,
+      temperature: 
+        payload.temperature !== undefined 
+        ? temperature / 2
+        : undefined,
       tools: buildAnthropicTools(tools),
       top_p,
     } satisfies Anthropic.MessageCreateParams;

diff --git a/src/libs/agent-runtime/bedrock/index.test.ts b/src/libs/agent-runtime/bedrock/index.test.ts
@@ -192,7 +192,7 @@ describe('LobeBedrockAI', () => {
             anthropic_version: 'bedrock-2023-05-31',
             max_tokens: 2048,
             messages: [{ content: 'Hello', role: 'user' }],
-            temperature: 0.5,
+            temperature: 0.25,
             top_p: 1,
           }),
           contentType: 'application/json',
@@ -230,7 +230,7 @@ describe('LobeBedrockAI', () => {
             anthropic_version: 'bedrock-2023-05-31',
             max_tokens: 2048,
             messages: [{ content: 'Hello', role: 'user' }],
-            temperature: 0.5,
+            temperature: 0.25,
             top_p: 1,
           }),
           contentType: 'application/json',

diff --git a/src/libs/agent-runtime/bedrock/index.ts b/src/libs/agent-runtime/bedrock/index.ts
@@ -64,7 +64,7 @@ export class LobeBedrockAI implements LobeRuntimeAI {
         max_tokens: max_tokens || 4096,
         messages: buildAnthropicMessages(user_messages),
         system: system_message?.content as string,
-        temperature: temperature,
+        temperature: temperature / 2,
         tools: buildAnthropicTools(tools),
         top_p: top_p,
       }),

diff --git a/src/libs/agent-runtime/groq/index.test.ts b/src/libs/agent-runtime/groq/index.test.ts
@@ -2,7 +2,7 @@
 import OpenAI from 'openai';
 import { Mock, afterEach, beforeEach, describe, expect, it, vi } from 'vitest';
 
-import { ChatStreamCallbacks, LobeOpenAICompatibleRuntime } from '@/libs/agent-runtime';
+import { LobeOpenAICompatibleRuntime } from '@/libs/agent-runtime';
 
 import * as debugStreamModule from '../utils/debugStream';
 import { LobeGroq } from './index';
@@ -318,3 +318,44 @@ describe('LobeGroqAI', () => {
     });
   });
 });
+
+describe('LobeGroqAI Temperature Tests', () => {
+  it('should set temperature to 0.7', async () => {
+    await instance.chat({
+      messages: [{ content: 'Hello', role: 'user' }],
+      model: 'mistralai/mistral-7b-instruct:free',
+      temperature: 0.7,
+    });
+
+    expect(instance['client'].chat.completions.create).toHaveBeenCalledWith(
+      expect.objectContaining({ temperature: 0.7 }),
+      expect.anything(),
+    );
+  });
+
+  it('should set temperature to 0', async () => {
+    await instance.chat({
+      messages: [{ content: 'Hello', role: 'user' }],
+      model: 'mistralai/mistral-7b-instruct:free',
+      temperature: 0,
+    });
+
+    expect(instance['client'].chat.completions.create).toHaveBeenCalledWith(
+      expect.objectContaining({ temperature: undefined }),
+      expect.anything(),
+    );
+  });
+
+  it('should set temperature to negative', async () => {
+    await instance.chat({
+      messages: [{ content: 'Hello', role: 'user' }],
+      model: 'mistralai/mistral-7b-instruct:free',
+      temperature: -1.0,
+    });
+
+    expect(instance['client'].chat.completions.create).toHaveBeenCalledWith(
+      expect.objectContaining({ temperature: undefined }),
+      expect.anything(),
+    );
+  });
+});
diff --git a/src/libs/agent-runtime/groq/index.ts b/src/libs/agent-runtime/groq/index.ts
@@ -11,10 +11,13 @@ export const LobeGroq = LobeOpenAICompatibleFactory({
         return { error, errorType: AgentRuntimeErrorType.LocationNotSupportError };
     },
     handlePayload: (payload) => {
+      const { temperature, ...restPayload } = payload;
       return {
-        ...payload,
+        ...restPayload,
         // disable stream for tools due to groq dont support
         stream: !payload.tools,
+
+        temperature: temperature <= 0 ? undefined : temperature,
       } as any;
     },
   },

diff --git a/src/libs/agent-runtime/minimax/index.test.ts b/src/libs/agent-runtime/minimax/index.test.ts
@@ -231,7 +231,7 @@ describe('LobeMinimaxAI', () => {
           messages: [{ content: 'Hello', role: 'user' }],
           model: 'text-davinci-003',
           stream: true,
-          temperature: 0.5,
+          temperature: 0.25,
           top_p: 0.8,
         });
       });

diff --git a/src/libs/agent-runtime/minimax/index.ts b/src/libs/agent-runtime/minimax/index.ts
@@ -141,7 +141,10 @@ export class LobeMinimaxAI implements LobeRuntimeAI {
       ...params,
       max_tokens: this.getMaxTokens(payload.model),
       stream: true,
-      temperature: temperature === 0 ? undefined : temperature,
+      temperature: 
+            temperature === undefined || temperature <= 0
+            ? undefined
+            : temperature / 2,
 
       tools: params.tools?.map((tool) => ({
         function: {

diff --git a/src/libs/agent-runtime/mistral/index.test.ts b/src/libs/agent-runtime/mistral/index.test.ts
@@ -81,7 +81,7 @@ describe('LobeMistralAI', () => {
           messages: [{ content: 'Hello', role: 'user' }],
           model: 'open-mistral-7b',
           stream: true,
-          temperature: 0.7,
+          temperature: 0.35,
           top_p: 1,
         },
         { headers: { Accept: '*/*' } },
@@ -114,7 +114,7 @@ describe('LobeMistralAI', () => {
           messages: [{ content: 'Hello', role: 'user' }],
           model: 'open-mistral-7b',
           stream: true,
-          temperature: 0.7,
+          temperature: 0.35,
           top_p: 1,
         },
         { headers: { Accept: '*/*' } },

diff --git a/src/libs/agent-runtime/mistral/index.ts b/src/libs/agent-runtime/mistral/index.ts
@@ -9,7 +9,10 @@ export const LobeMistralAI = LobeOpenAICompatibleFactory({
       messages: payload.messages as any,
       model: payload.model,
       stream: true,
-      temperature: payload.temperature,
+      temperature: 
+        payload.temperature !== undefined 
+        ? payload.temperature / 2
+        : undefined,
       ...payload.tools && { tools: payload.tools },
       top_p: payload.top_p,
     }),

diff --git a/src/libs/agent-runtime/moonshot/index.ts b/src/libs/agent-runtime/moonshot/index.ts
@@ -1,8 +1,23 @@
-import { ModelProvider } from '../types';
+import OpenAI from 'openai';
+
+import { ChatStreamPayload, ModelProvider } from '../types';
 import { LobeOpenAICompatibleFactory } from '../utils/openaiCompatibleFactory';
 
 export const LobeMoonshotAI = LobeOpenAICompatibleFactory({
   baseURL: 'https://api.moonshot.cn/v1',
+  chatCompletion: {
+    handlePayload: (payload: ChatStreamPayload) => {
+      const { temperature, ...rest } = payload;
+
+      return { 
+        ...rest, 
+        temperature: 
+          temperature !== undefined 
+          ? temperature / 2
+          : undefined,
+      } as OpenAI.ChatCompletionCreateParamsStreaming;
+    },
+  },
   debug: {
     chatCompletion: () => process.env.DEBUG_MOONSHOT_CHAT_COMPLETION === '1',
   },

diff --git a/src/libs/agent-runtime/ollama/index.test.ts b/src/libs/agent-runtime/ollama/index.test.ts
@@ -103,6 +103,33 @@ describe('LobeOllamaAI', () => {
 
       expect(abortMock).toHaveBeenCalled();
     });
+
+    it('temperature should be divided by two', async () => {
+      const chatMock = vi.fn().mockResolvedValue({});
+      vi.mocked(Ollama.prototype.chat).mockImplementation(chatMock);
+
+      const payload = {
+        messages: [{ content: 'Hello', role: 'user' }],
+        model: 'model-id',
+        temperature: 0.7,
+      };
+      const options = { signal: new AbortController().signal };
+
+      const response = await ollamaAI.chat(payload as any, options);
+
+      expect(chatMock).toHaveBeenCalledWith({
+        messages: [{ content: 'Hello', role: 'user' }],
+        model: 'model-id',
+        options: {
+          frequency_penalty: undefined,
+          presence_penalty: undefined,
+          temperature: 0.35,
+          top_p: undefined,
+        },
+        stream: true,
+      });
+      expect(response).toBeInstanceOf(Response);
+    });
   });
 
   describe('models', () => {

diff --git a/src/libs/agent-runtime/ollama/index.ts b/src/libs/agent-runtime/ollama/index.ts
@@ -45,7 +45,10 @@ export class LobeOllamaAI implements LobeRuntimeAI {
         options: {
           frequency_penalty: payload.frequency_penalty,
           presence_penalty: payload.presence_penalty,
-          temperature: payload.temperature,
+          temperature: 
+            payload.temperature !== undefined 
+            ? payload.temperature / 2
+            : undefined,
           top_p: payload.top_p,
         },
         stream: true,

diff --git a/src/libs/agent-runtime/perplexity/index.test.ts b/src/libs/agent-runtime/perplexity/index.test.ts
@@ -245,5 +245,47 @@ describe('LobePerplexityAI', () => {
         process.env.DEBUG_PERPLEXITY_CHAT_COMPLETION = originalDebugValue;
       });
     });
+
+    it('should call chat method with temperature', async () => {
+      vi.spyOn(instance['client'].chat.completions, 'create').mockResolvedValue(
+        new ReadableStream() as any,
+      );
+
+      await instance.chat({
+        messages: [{ content: 'Hello', role: 'user' }],
+        model: 'text-davinci-003',
+        temperature: 1.5,
+      });
+
+      expect(instance['client'].chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          messages: expect.any(Array),
+          model: 'text-davinci-003',
+          temperature: 1.5,
+        }),
+        expect.any(Object),
+      );
+    });
+
+    it('should be undefined when temperature >= 2', async () => {
+      vi.spyOn(instance['client'].chat.completions, 'create').mockResolvedValue(
+        new ReadableStream() as any,
+      );
+
+      await instance.chat({
+        messages: [{ content: 'Hello', role: 'user' }],
+        model: 'text-davinci-003',
+        temperature: 2,
+      });
+
+      expect(instance['client'].chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          messages: expect.any(Array),
+          model: 'text-davinci-003',
+          temperature: undefined,
+        }),
+        expect.any(Object),
+      );
+    });
   });
 });
diff --git a/src/libs/agent-runtime/perplexity/index.ts b/src/libs/agent-runtime/perplexity/index.ts
@@ -8,7 +8,7 @@ export const LobePerplexityAI = LobeOpenAICompatibleFactory({
   chatCompletion: {
     handlePayload: (payload: ChatStreamPayload) => {
       // Set a default frequency penalty value greater than 0
-      const { presence_penalty, frequency_penalty, stream = true, ...res } = payload;
+      const { presence_penalty, frequency_penalty, stream = true, temperature,  ...res } = payload;
 
       let param;
 
@@ -21,7 +21,12 @@ export const LobePerplexityAI = LobeOpenAICompatibleFactory({
         param = { frequency_penalty: frequency_penalty || defaultFrequencyPenalty };
       }
 
-      return { ...res, ...param, stream } as OpenAI.ChatCompletionCreateParamsStreaming;
+      return { 
+        ...res, 
+        ...param, 
+        stream,
+        temperature: temperature >= 2 ? undefined : temperature,
+      } as OpenAI.ChatCompletionCreateParamsStreaming;
     },
   },
   debug: {