Gianluca Sabato

Results 5 comments of Gianluca Sabato

When using RAG or Multi-Agent features with llama models the script stops replying without meeting the termination condition or errors

Hi and thanks, @MohMaz. I’ve tried the following provisioning modes: - **Ollama (local)** - **Ollama with liteLLM (local)** - **together.ai (remote)** I’ve been focusing on various Llama versions and tested...

When using RAG or Multi-Agent features with llama models the script stops replying without meeting the termination condition or errors

It happened even without truncation. I added it trying to solve the problem. I'll post later an output without truncation. Or you can try it disabling the truncation in the...

When using RAG or Multi-Agent features with llama models the script stops replying without meeting the termination condition or errors

Here it's an output without truncation, the same problem happens. ``` ============================== Starting Chat using model: meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo ============================== Trying to create collection. 2024-11-29 21:45:19,089 - autogen.agentchat.contrib.retrieve_user_proxy_agent - INFO - Use...

When using RAG or Multi-Agent features with llama models the script stops replying without meeting the termination condition or errors

@thinkall Hi and thanks. I didn't try this exact code iteration with OpenAI. In general, the same code performed better and in a most reliable way with OpenAI APIs. I...

When using RAG or Multi-Agent features with llama models the script stops replying without meeting the termination condition or errors

@thinkall I tried to fix the `custom_speaker_selection_func` function like this: ``` def custom_speaker_selection_func(last_speaker: Agent, groupchat: GroupChat): """Define a customized speaker selection function. A recommended way is to define a transition...