[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (8.16.0) #199303
Assignees
Labels
bug
Fixes for quality problems that affect the customer experience
fixed
Team:Security Generative AI
Security Generative AI
Team: SecuritySolution
Security Solutions Team working on SIEM, Endpoint, Timeline, Resolver, etc.
Comments
mick-lue
added
bug
|
Pinging @elastic/security-solution (Team: SecuritySolution) |
|
Update: The issue still persists. When creating a new chat via the "New chat" button, this chat will default to using the model gpt-4o in its requests. After going into the chat options (that are now not on the same page anymore) and saving the chat again without modifying the connector, it correctly loads the model that is specified in that connector. |
mick-lue
changed the title
e40pud
added a commit
to e40pud/kibana
that referenced
this issue
Dec 12, 2024
…w chat does not use connector's model (elastic#199303)
e40pud
added
Team:Security Generative AI
e40pud
added a commit
that referenced
this issue
Dec 14, 2024
…w chat does not use connector's model (#199303) (#204014) ## Summary The PR fixes [this bug](#199303) The issue happens with some of the locally setup LLMs (like [Ollama](https://github.com/ollama/ollama)) which requires the correct `model` to be passed as part of the [chat completions API](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion). We had a bug in our code when on new conversation creation we would not pass all the connectors configuration and only `connectorId` and `actionTypeId` would be passed. Here is the old code implementation: ``` const newConversation = await createConversation({ title: NEW_CHAT, ...(currentConversation?.apiConfig != null && currentConversation?.apiConfig?.actionTypeId != null ? { apiConfig: { connectorId: currentConversation.apiConfig.connectorId, actionTypeId: currentConversation.apiConfig.actionTypeId, ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}), }, } : {}), }); ``` and thus the new conversation would not have the complete connector configuration which would cause to use default model (`gpt-4o`) as a model we pass to the LLM. Also, I updated the default body that we use on the Test connector page, to make sure that we send a model parameter to the LLM in case of `Open AI > Other (OpenAI Compatible Service)` kind of connectors. ### Testing notes Steps to reproduce: 1. Install [Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama) locally 2. Setup an OpenAI connector using Other (OpenAI Compatible Service) provider 3. Open AI Assistant and select created Ollama connector to chat 4. Create a "New Chat" 5. The Ollama connector should be selected 6. Send a message to LLM (for example "hello world") Expected: there should be no errors saying `ActionsClientChatOpenAI: an error occurred while running the action - Unexpected API Error: - 404 model "gpt-4o" not found, try pulling it first`
kibanamachine
pushed a commit
to kibanamachine/kibana
that referenced
this issue
Dec 14, 2024
…w chat does not use connector's model (elastic#199303) (elastic#204014) ## Summary The PR fixes [this bug](elastic#199303) The issue happens with some of the locally setup LLMs (like [Ollama](https://github.com/ollama/ollama)) which requires the correct `model` to be passed as part of the [chat completions API](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion). We had a bug in our code when on new conversation creation we would not pass all the connectors configuration and only `connectorId` and `actionTypeId` would be passed. Here is the old code implementation: ``` const newConversation = await createConversation({ title: NEW_CHAT, ...(currentConversation?.apiConfig != null && currentConversation?.apiConfig?.actionTypeId != null ? { apiConfig: { connectorId: currentConversation.apiConfig.connectorId, actionTypeId: currentConversation.apiConfig.actionTypeId, ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}), }, } : {}), }); ``` and thus the new conversation would not have the complete connector configuration which would cause to use default model (`gpt-4o`) as a model we pass to the LLM. Also, I updated the default body that we use on the Test connector page, to make sure that we send a model parameter to the LLM in case of `Open AI > Other (OpenAI Compatible Service)` kind of connectors. ### Testing notes Steps to reproduce: 1. Install [Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama) locally 2. Setup an OpenAI connector using Other (OpenAI Compatible Service) provider 3. Open AI Assistant and select created Ollama connector to chat 4. Create a "New Chat" 5. The Ollama connector should be selected 6. Send a message to LLM (for example "hello world") Expected: there should be no errors saying `ActionsClientChatOpenAI: an error occurred while running the action - Unexpected API Error: - 404 model "gpt-4o" not found, try pulling it first` (cherry picked from commit 7e4e859)
kibanamachine
pushed a commit
to kibanamachine/kibana
that referenced
this issue
Dec 14, 2024
…w chat does not use connector's model (elastic#199303) (elastic#204014) ## Summary The PR fixes [this bug](elastic#199303) The issue happens with some of the locally setup LLMs (like [Ollama](https://github.com/ollama/ollama)) which requires the correct `model` to be passed as part of the [chat completions API](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion). We had a bug in our code when on new conversation creation we would not pass all the connectors configuration and only `connectorId` and `actionTypeId` would be passed. Here is the old code implementation: ``` const newConversation = await createConversation({ title: NEW_CHAT, ...(currentConversation?.apiConfig != null && currentConversation?.apiConfig?.actionTypeId != null ? { apiConfig: { connectorId: currentConversation.apiConfig.connectorId, actionTypeId: currentConversation.apiConfig.actionTypeId, ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}), }, } : {}), }); ``` and thus the new conversation would not have the complete connector configuration which would cause to use default model (`gpt-4o`) as a model we pass to the LLM. Also, I updated the default body that we use on the Test connector page, to make sure that we send a model parameter to the LLM in case of `Open AI > Other (OpenAI Compatible Service)` kind of connectors. ### Testing notes Steps to reproduce: 1. Install [Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama) locally 2. Setup an OpenAI connector using Other (OpenAI Compatible Service) provider 3. Open AI Assistant and select created Ollama connector to chat 4. Create a "New Chat" 5. The Ollama connector should be selected 6. Send a message to LLM (for example "hello world") Expected: there should be no errors saying `ActionsClientChatOpenAI: an error occurred while running the action - Unexpected API Error: - 404 model "gpt-4o" not found, try pulling it first` (cherry picked from commit 7e4e859)
kibanamachine
pushed a commit
to kibanamachine/kibana
that referenced
this issue
Dec 14, 2024
…w chat does not use connector's model (elastic#199303) (elastic#204014) ## Summary The PR fixes [this bug](elastic#199303) The issue happens with some of the locally setup LLMs (like [Ollama](https://github.com/ollama/ollama)) which requires the correct `model` to be passed as part of the [chat completions API](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion). We had a bug in our code when on new conversation creation we would not pass all the connectors configuration and only `connectorId` and `actionTypeId` would be passed. Here is the old code implementation: ``` const newConversation = await createConversation({ title: NEW_CHAT, ...(currentConversation?.apiConfig != null && currentConversation?.apiConfig?.actionTypeId != null ? { apiConfig: { connectorId: currentConversation.apiConfig.connectorId, actionTypeId: currentConversation.apiConfig.actionTypeId, ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}), }, } : {}), }); ``` and thus the new conversation would not have the complete connector configuration which would cause to use default model (`gpt-4o`) as a model we pass to the LLM. Also, I updated the default body that we use on the Test connector page, to make sure that we send a model parameter to the LLM in case of `Open AI > Other (OpenAI Compatible Service)` kind of connectors. ### Testing notes Steps to reproduce: 1. Install [Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama) locally 2. Setup an OpenAI connector using Other (OpenAI Compatible Service) provider 3. Open AI Assistant and select created Ollama connector to chat 4. Create a "New Chat" 5. The Ollama connector should be selected 6. Send a message to LLM (for example "hello world") Expected: there should be no errors saying `ActionsClientChatOpenAI: an error occurred while running the action - Unexpected API Error: - 404 model "gpt-4o" not found, try pulling it first` (cherry picked from commit 7e4e859)
kibanamachine
added a commit
that referenced
this issue
Dec 14, 2024
…bug. New chat does not use connector's model (#199303) (#204014) (#204306) # Backport This will backport the following commits from `main` to `8.16`: - [[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)](#204014) <!--- Backport version: 9.4.3 --> ### Questions ? Please refer to the [Backport tool documentation](https://github.com/sqren/backport) <!--BACKPORT [{"author":{"name":"Ievgen Sorokopud","email":"[email protected]"},"sourceCommit":{"committedDate":"2024-12-14T08:54:54Z","message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0","branchLabelMapping":{"^v9.0.0$":"main","^v8.18.0$":"8.x","^v(\\d+).(\\d+).\\d+$":"$1.$2"}},"sourcePullRequest":{"labels":["release_note:fix","v9.0.0","Team: SecuritySolution","Team:Security Generative AI","backport:version","v8.18.0","v8.16.3","v8.17.1"],"title":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303)","number":204014,"url":"https://github.com/elastic/kibana/pull/204014","mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},"sourceBranch":"main","suggestedTargetBranches":["8.x","8.16","8.17"],"targetPullRequestStates":[{"branch":"main","label":"v9.0.0","branchLabelMappingKey":"^v9.0.0$","isSourceBranch":true,"state":"MERGED","url":"https://github.com/elastic/kibana/pull/204014","number":204014,"mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},{"branch":"8.x","label":"v8.18.0","branchLabelMappingKey":"^v8.18.0$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.16","label":"v8.16.3","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.17","label":"v8.17.1","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"}]}] BACKPORT--> Co-authored-by: Ievgen Sorokopud <[email protected]>
kibanamachine
added a commit
that referenced
this issue
Dec 14, 2024
…ug. New chat does not use connector's model (#199303) (#204014) (#204308) # Backport This will backport the following commits from `main` to `8.x`: - [[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)](#204014) <!--- Backport version: 9.4.3 --> ### Questions ? Please refer to the [Backport tool documentation](https://github.com/sqren/backport) <!--BACKPORT [{"author":{"name":"Ievgen Sorokopud","email":"[email protected]"},"sourceCommit":{"committedDate":"2024-12-14T08:54:54Z","message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0","branchLabelMapping":{"^v9.0.0$":"main","^v8.18.0$":"8.x","^v(\\d+).(\\d+).\\d+$":"$1.$2"}},"sourcePullRequest":{"labels":["release_note:fix","v9.0.0","Team: SecuritySolution","Team:Security Generative AI","backport:version","v8.18.0","v8.16.3","v8.17.1"],"title":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303)","number":204014,"url":"https://github.com/elastic/kibana/pull/204014","mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},"sourceBranch":"main","suggestedTargetBranches":["8.x","8.16","8.17"],"targetPullRequestStates":[{"branch":"main","label":"v9.0.0","branchLabelMappingKey":"^v9.0.0$","isSourceBranch":true,"state":"MERGED","url":"https://github.com/elastic/kibana/pull/204014","number":204014,"mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},{"branch":"8.x","label":"v8.18.0","branchLabelMappingKey":"^v8.18.0$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.16","label":"v8.16.3","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.17","label":"v8.17.1","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"}]}] BACKPORT--> Co-authored-by: Ievgen Sorokopud <[email protected]>
kibanamachine
added a commit
that referenced
this issue
Dec 14, 2024
…bug. New chat does not use connector's model (#199303) (#204014) (#204307) # Backport This will backport the following commits from `main` to `8.17`: - [[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)](#204014) <!--- Backport version: 9.4.3 --> ### Questions ? Please refer to the [Backport tool documentation](https://github.com/sqren/backport) <!--BACKPORT [{"author":{"name":"Ievgen Sorokopud","email":"[email protected]"},"sourceCommit":{"committedDate":"2024-12-14T08:54:54Z","message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0","branchLabelMapping":{"^v9.0.0$":"main","^v8.18.0$":"8.x","^v(\\d+).(\\d+).\\d+$":"$1.$2"}},"sourcePullRequest":{"labels":["release_note:fix","v9.0.0","Team: SecuritySolution","Team:Security Generative AI","backport:version","v8.18.0","v8.16.3","v8.17.1"],"title":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303)","number":204014,"url":"https://github.com/elastic/kibana/pull/204014","mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},"sourceBranch":"main","suggestedTargetBranches":["8.x","8.16","8.17"],"targetPullRequestStates":[{"branch":"main","label":"v9.0.0","branchLabelMappingKey":"^v9.0.0$","isSourceBranch":true,"state":"MERGED","url":"https://github.com/elastic/kibana/pull/204014","number":204014,"mergeCommit":{"message":"[Security Solution] AI Assistant: LLM Connector model chooser bug. New chat does not use connector's model (#199303) (#204014)\n\n## Summary\r\n\r\nThe PR fixes [this bug](https://github.com/elastic/kibana/issues/199303)\r\n\r\nThe issue happens with some of the locally setup LLMs (like\r\n[Ollama](https://github.com/ollama/ollama)) which requires the correct\r\n`model` to be passed as part of the [chat completions\r\nAPI](https://github.com/ollama/ollama/blob/main/docs/api.md#generate-a-chat-completion).\r\n\r\nWe had a bug in our code when on new conversation creation we would not\r\npass all the connectors configuration and only `connectorId` and\r\n`actionTypeId` would be passed. Here is the old code implementation:\r\n\r\n```\r\nconst newConversation = await createConversation({\r\n title: NEW_CHAT,\r\n ...(currentConversation?.apiConfig != null &&\r\n currentConversation?.apiConfig?.actionTypeId != null\r\n ? {\r\n apiConfig: {\r\n connectorId: currentConversation.apiConfig.connectorId,\r\n actionTypeId: currentConversation.apiConfig.actionTypeId,\r\n ...(newSystemPrompt?.id != null ? { defaultSystemPromptId: newSystemPrompt.id } : {}),\r\n },\r\n }\r\n : {}),\r\n});\r\n```\r\n\r\nand thus the new conversation would not have the complete connector\r\nconfiguration which would cause to use default model (`gpt-4o`) as a\r\nmodel we pass to the LLM.\r\n\r\nAlso, I updated the default body that we use on the Test connector page,\r\nto make sure that we send a model parameter to the LLM in case of `Open\r\nAI > Other (OpenAI Compatible Service)` kind of connectors.\r\n\r\n### Testing notes\r\n\r\nSteps to reproduce:\r\n1. Install\r\n[Ollama](https://github.com/ollama/ollama?tab=readme-ov-file#ollama)\r\nlocally\r\n2. Setup an OpenAI connector using Other (OpenAI Compatible Service)\r\nprovider\r\n3. Open AI Assistant and select created Ollama connector to chat\r\n4. Create a \"New Chat\"\r\n5. The Ollama connector should be selected\r\n6. Send a message to LLM (for example \"hello world\")\r\n\r\nExpected: there should be no errors saying `ActionsClientChatOpenAI: an\r\nerror occurred while running the action - Unexpected API Error: - 404\r\nmodel \"gpt-4o\" not found, try pulling it first`","sha":"7e4e8592f45ceca822c4f34d18e9f047cfe3cde0"}},{"branch":"8.x","label":"v8.18.0","branchLabelMappingKey":"^v8.18.0$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.16","label":"v8.16.3","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"},{"branch":"8.17","label":"v8.17.1","branchLabelMappingKey":"^v(\\d+).(\\d+).\\d+$","isSourceBranch":false,"state":"NOT_CREATED"}]}] BACKPORT--> Co-authored-by: Ievgen Sorokopud <[email protected]>
|
@MadameSheema this issue was fixed and merged. Fix will be available in next versions: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug:
When opening a new chat, the LLM Connector's default model is not used.
Additionally, the model list only shows pre-configured options and does not show the default model or other added model names. When entering a previously used model name though, it errors and resets to the global default (in this case "gpt-4o") instead of choosing the already existing model name.
Kibana/Elasticsearch Stack version:
8.15.3 (self-hosted)
Server OS version:
Ubuntu 22.04.5 LTS
Browser and Browser OS versions:
Vivaldi for Windows (7.0.3495.10 (Stable channel) (64-Bit))
Original install method (e.g. download page, yum, from source, etc.):
apt for Elastic Stack
BYO LLM from https://www.elastic.co/guide/en/security/8.15/connect-to-byo-llm.html
Functional Area (e.g. Endpoint management, timelines, resolver, etc.):
AI Assistant (in Security Solution context)
using OpenAI Connector (OpenAI provider)
Steps to reproduce:
Current behavior:
The new chat will be configured with the previously chosen chat's model name.
The model name can't be changed to an "already existing" value. An error might occur and the model will default to "gpt-4o" (seen in the request, but not in the UI).
Expected behavior:
The new chat should be configured with the Connector's default model value (OR maybe a chosen Connector should persist the chosen model through all chat tabs).
The configured model names should be listed in the dropdown menu. A previously configured model name should be selectable and not result in an error.
Screenshots (if relevant):
The text was updated successfully, but these errors were encountered: