Files
pr-agent/docs/docs/usage-guide/changing_a_model.md
2025-02-26 10:15:22 +02:00

224 lines
8.5 KiB
Markdown

## Changing a model in PR-Agent
See [here](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/algo/__init__.py) for a list of available models.
To use a different model than the default (GPT-4), you need to edit in the [configuration file](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/settings/configuration.toml#L2) the fields:
```
[config]
model = "..."
fallback_models = ["..."]
```
For models and environments not from OpenAI, you might need to provide additional keys and other parameters.
You can give parameters via a configuration file (see below for instructions), or from environment variables. See [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms) for the environment variables relevant per model.
### Azure
To use Azure, set in your `.secrets.toml` (working from CLI), or in the GitHub `Settings > Secrets and variables` (working from GitHub App or GitHub Action):
```
[openai]
key = "" # your azure api key
api_type = "azure"
api_version = '2023-05-15' # Check Azure documentation for the current API version
api_base = "" # The base URL for your Azure OpenAI resource. e.g. "https://<your resource name>.openai.azure.com"
deployment_id = "" # The deployment name you chose when you deployed the engine
```
and set in your configuration file:
```
[config]
model="" # the OpenAI model you've deployed on Azure (e.g. gpt-4o)
fallback_models=["..."]
```
Passing custom headers to the underlying LLM Model API can be done by setting extra_headers parameter to litellm.
```
[litellm]
extra_headers='{"projectId": "<authorized projectId >", ...}') #The value of this setting should be a JSON string representing the desired headers, a ValueError is thrown otherwise.
```
This enables users to pass authorization tokens or API keys, when routing requests through an API management gateway.
### Ollama
You can run models locally through either [VLLM](https://docs.litellm.ai/docs/providers/vllm) or [Ollama](https://docs.litellm.ai/docs/providers/ollama)
E.g. to use a new model locally via Ollama, set in `.secrets.toml` or in a configuration file:
```
[config]
model = "ollama/qwen2.5-coder:32b"
fallback_models=["ollama/qwen2.5-coder:32b"]
custom_model_max_tokens=128000 # set the maximal input tokens for the model
duplicate_examples=true # will duplicate the examples in the prompt, to help the model to generate structured output
[ollama]
api_base = "http://localhost:11434" # or whatever port you're running Ollama on
```
!!! note "Local models vs commercial models"
Qodo Merge is compatible with almost any AI model, but analyzing complex code repositories and pull requests requires a model specifically optimized for code analysis.
Commercial models such as GPT-4, Claude Sonnet, and Gemini have demonstrated robust capabilities in generating structured output for code analysis tasks with large input. In contrast, most open-source models currently available (as of January 2025) face challenges with these complex tasks.
Based on our testing, local open-source models are suitable for experimentation and learning purposes (mainly for the `ask` command), but they are not suitable for production-level code analysis tasks.
Hence, for production workflows and real-world usage, we recommend using commercial models.
### Hugging Face
To use a new model with Hugging Face Inference Endpoints, for example, set:
```
[config] # in configuration.toml
model = "huggingface/meta-llama/Llama-2-7b-chat-hf"
fallback_models=["huggingface/meta-llama/Llama-2-7b-chat-hf"]
custom_model_max_tokens=... # set the maximal input tokens for the model
[huggingface] # in .secrets.toml
key = ... # your Hugging Face api key
api_base = ... # the base url for your Hugging Face inference endpoint
```
(you can obtain a Llama2 key from [here](https://replicate.com/replicate/llama-2-70b-chat/api))
### Replicate
To use Llama2 model with Replicate, for example, set:
```
[config] # in configuration.toml
model = "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
fallback_models=["replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"]
[replicate] # in .secrets.toml
key = ...
```
(you can obtain a Llama2 key from [here](https://replicate.com/replicate/llama-2-70b-chat/api))
Also, review the [AiHandler](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/algo/ai_handler.py) file for instructions on how to set keys for other models.
### Groq
To use Llama3 model with Groq, for example, set:
```
[config] # in configuration.toml
model = "llama3-70b-8192"
fallback_models = ["groq/llama3-70b-8192"]
[groq] # in .secrets.toml
key = ... # your Groq api key
```
(you can obtain a Groq key from [here](https://console.groq.com/keys))
### Vertex AI
To use Google's Vertex AI platform and its associated models (chat-bison/codechat-bison) set:
```
[config] # in configuration.toml
model = "vertex_ai/codechat-bison"
fallback_models="vertex_ai/codechat-bison"
[vertexai] # in .secrets.toml
vertex_project = "my-google-cloud-project"
vertex_location = ""
```
Your [application default credentials](https://cloud.google.com/docs/authentication/application-default-credentials) will be used for authentication so there is no need to set explicit credentials in most environments.
If you do want to set explicit credentials, then you can use the `GOOGLE_APPLICATION_CREDENTIALS` environment variable set to a path to a json credentials file.
### Google AI Studio
To use [Google AI Studio](https://aistudio.google.com/) models, set the relevant models in the configuration section of the configuration file:
```toml
[config] # in configuration.toml
model="google_ai_studio/gemini-1.5-flash"
fallback_models=["google_ai_studio/gemini-1.5-flash"]
[google_ai_studio] # in .secrets.toml
gemini_api_key = "..."
```
If you don't want to set the API key in the .secrets.toml file, you can set the `GOOGLE_AI_STUDIO.GEMINI_API_KEY` environment variable.
### Anthropic
To use Anthropic models, set the relevant models in the configuration section of the configuration file:
```
[config]
model="anthropic/claude-3-opus-20240229"
fallback_models=["anthropic/claude-3-opus-20240229"]
```
And also set the api key in the .secrets.toml file:
```
[anthropic]
KEY = "..."
```
### Amazon Bedrock
To use Amazon Bedrock and its foundational models, add the below configuration:
```
[config] # in configuration.toml
model="bedrock/anthropic.claude-3-sonnet-20240229-v1:0"
fallback_models=["bedrock/anthropic.claude-v2:1"]
```
Note that you have to add access to foundational models before using them. Please refer to [this document](https://docs.aws.amazon.com/bedrock/latest/userguide/setting-up.html) for more details.
If you are using the claude-3 model, please configure the following settings as there are parameters incompatible with claude-3.
```
[litellm]
drop_params = true
```
AWS session is automatically authenticated from your environment, but you can also explicitly set `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY` and `AWS_REGION_NAME` environment variables. Please refer to [this document](https://litellm.vercel.app/docs/providers/bedrock) for more details.
### DeepSeek
To use deepseek-chat model with DeepSeek, for example, set:
```toml
[config] # in configuration.toml
model = "deepseek/deepseek-chat"
fallback_models=["deepseek/deepseek-chat"]
```
and fill up your key
```toml
[deepseek] # in .secrets.toml
key = ...
```
(you can obtain a deepseek-chat key from [here](https://platform.deepseek.com))
### Custom models
If the relevant model doesn't appear [here](https://github.com/Codium-ai/pr-agent/blob/main/pr_agent/algo/__init__.py), you can still use it as a custom model:
1. Set the model name in the configuration file:
```
[config]
model="custom_model_name"
fallback_models=["custom_model_name"]
```
2. Set the maximal tokens for the model:
```
[config]
custom_model_max_tokens= ...
```
3. Go to [litellm documentation](https://litellm.vercel.app/docs/proxy/quick_start#supported-llms), find the model you want to use, and set the relevant environment variables.
4. Most reasoning models do not support chat-style inputs (`system` and `user` messages) or temperature settings.
To bypass chat templates and temperature controls, set `config.custom_reasoning_model = true` in your configuration file.
## Dedicated parameters
### OpenAI models
[config]
reasoning_efffort= = "medium" # "low", "medium", "high"
With the OpenAI models that support reasoning effort (eg: o3-mini), you can specify its reasoning effort via `config` section. The default value is `medium`. You can change it to `high` or `low` based on your usage.