You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
AndroidStudioSettings/ChatGPTConversations_170.xml

6 lines
37 KiB

<application>
<component name="ee.carlrobert.codegpt.state.conversations.ConversationsState">
<option name="currentConversation" value="{&quot;id&quot;:&quot;e229fb79-598c-41af-8261-de0a81b167d2&quot;,&quot;title&quot;:null,&quot;messages&quot;:[{&quot;prompt&quot;:&quot;我注册了Fireworks怎么确定它提供的哪个模型支持FIM&quot;,&quot;id&quot;:&quot;46ff0f61-ea31-49d9-88f3-8dc21f9133f2&quot;,&quot;response&quot;:&quot;提供的代码上下文与您的问题无关。关于如何在 Fireworks AI FIM (Fill-In-the-Middle/)\n\n### 1. (Code Models)\nFIM Fireworks [Models ](https://fireworks.ai/models) FIM \n* **Qwen-Coder** `qwen2.5-coder-32b` \n* **StarCoder / StarCoder2** \n* **CodeLlama** \n* **DeepSeek-Coder** \n\n### 2. (Model Card) \n Fireworks FIM\n* `FIM` `Fill-in-the-middle`\n* `Code completion` ()\n* `Infilling`\n\n### 3. Base () \n Qwen2.5-Coder **Base ** FIM **Instruct ** Instruct FIM Base FIM \n\n### 4. Token API \n使 Token FIM 怀 FIM Fireworks Completions API `/v1/completions` `/v1/chat/completions` FIM Token Prompt \n\nQwen StarCoder 使 Token \n```text\n&lt;|fim_prefix|&gt;&lt;|fim_suffix|&gt;&lt;|fim_middle|&gt;\n```\n\n****\n```json\n{\n \&quot;model\&quot;: \&quot;accounts/fireworks/models/qwen2p5-coder-32b\&quot;,\n \&quot;prompt\&quot;: \&quot;&lt;|fim_prefix|&gt;def calculate_area(radius):\\n &lt;|fim_suffix|&gt;\\n return area&lt;|fim_middle|&gt;\&quot;,\n \&quot;max_tokens\&quot;: 50,\n \&quot;temperature\&quot;: 0.2\n}\n```\n FIM `&lt;|fim_middle|&gt;` `area = 3.14 * radius ** 2` &quot;,&quot;referencedFilePaths&quot;:null,&quot;conversationsHistoryIds&quot;:null,&quot;imageFilePath&quot;:null,&quot;webSearchIncluded&quot;:false,&quot;personaName&quot;:null,&quot;tool_calls&quot;:null,&quot;tool_call_results&quot;:null},{&quot;prompt&quot;:&quot;Qwen2.5-Coder Qwen3-Coder &quot;,&quot;id&quot;:&quot;ad74c041-a5f6-427b-b92d-18f335e08704&quot;,&quot;response&quot;:&quot;**Qwen3-Coder ** **Qwen2.5-Coder** 2024 \n\n **Qwen2.5-Coder** **Qwen3-Coder** \n\n### Qwen2.5-Coder\nQwen2.5-Coder \n1. **** 0.5B1.5B3B7B14B32B 32B HumanEvalMBPP GPT-4o\n2. ** FIM ** Base Fill-in-the-Middle IDE \n3. **** 32K <EFBFBD>
<option name="conversations" value="[{&quot;id&quot;:&quot;2ea372bf-f79d-4805-a228-8cd51bf07dc9&quot;,&quot;title&quot;:null,&quot;messages&quot;:[{&quot;prompt&quot;:&quot;我在适用ProxyAI这个Intellij插件想要实现代码自动补全gemini系列的哪个模型最适合为什么另外配置里有个“FIM”模板选项我应该选什么&quot;,&quot;id&quot;:&quot;ab4b92a6-8055-4844-85d6-57ffd6339399&quot;,&quot;response&quot;:&quot;对于在 IntelliJ IDEA 使 ProxyAI **Inline Code Completion** Gemini FIM \n\n### Gemini 1.5 Flash Gemini 2.0 Flash\n\n**使 `gemini-1.5-flash` `gemini-2.0-flash`** API \n\n** Pro **\n1. **Latency** 300-500 `gemini-1.5-pro` Flash \n2. ****Flash 1M 2M ProxyAI \n3. ****Flash API Pro \n\n* `gemini-1.5-pro`****Flash *\n\n---\n\n### FIM \n\n**FIM (Fill-In-the-Middle)** PrefixSuffixMiddle\n\n**Gemini **\n DeepSeek CoderQwen Coder CodeLlama 使 FIM Token `&lt;|fim_prefix|&gt;``&lt;|fim_suffix|&gt;` **Gemini FIM ** ProxyAI Gemini `DeepSeek` `Qwen` FIM Gemini \n\n****\n\n1. ** Prompt / Instruction **\n ProxyAI FIM `Prompt``Instruction` `Gemini/OpenAI` \n\n2. **CustomFIM **\n FIM 使 XML Gemini XML \n ```text\n You are an expert programmer. Please complete the missing code between &lt;prefix&gt; and &lt;suffix&gt;. \n Output ONLY the missing code, without any markdown formatting, explanations, or repeating the prefix/suffix.\n \n &lt;prefix&gt;\n {prefix}\n &lt;/prefix&gt;\n &lt;suffix&gt;\n {suffix}\n &lt;/suffix&gt;\n ```\n *( `{prefix}` `{suffix}` ProxyAI )*\n\n3. ** FIM**\n FIM Gemini ProxyAI ** FIM ** None/Default退 Gemini &quot;,&quot;referencedFilePaths&quot;:null,&quot;conversationsHistoryIds&quot;:null,&quot;imageFilePath&quot;:null,&quot;
</component>
</application>