Video

How to disable telemetry? (Data collection)

Hello, please make a review about Open WebUI. It's the great hype currently. "Open WebUI", is a frontend for LLMs, which offers the ability for users to talk hands free, with excellent voices, using internal or external TTS APIs from Eleven Labs, Groq, etc, has web search, RAG, long term memory, it's compatible with OpenAI API, can import GGUF models directly from Hugging Face and convert for using on Ollama, can fine tune models for special needs, can work with multimodal models, image and video, can engage with multiple models simultaneously, local and external at the same time, can accept new plugins, external tools and functions... And it's also open source. 🎉❤

Now we can with GPT 4o mini. The prices are 90% cheaper ! ❤

🇧🇷🇧🇷🇧🇷👏👏👏

9:35 Process finished with exit code 0. Not same as the tutorial. End of game. not printing nothing

Thank you brother! Also, no homo, you are a good looking man!

You're awesome, reviewing stuff for an interview on Monday, I so forgot all this s**t.

can it do anything more complex though?

Hi, I was wondering how what if the target node is 12? Since the algorithm never goes back? Thanks

Can the workflow implement graph? Though it feels clunkier to implement at its current state

Is there seriously no equivalent to Stability Matrix to handle all the package downloading and connecting for us? I couldn't even get Automatic1111 installed properly from commandlines, spending days banging my head against it before someone said "why don't you just use Stability Matrix?" For chatbots, there's GPT4All to automate the model handling. I was seriously hoping something similar existed for handling music AI packages. Install a wrapper, then click the stuff you think you want, and other than watching disk space vanish like fuel in a 1972 Barracuda, it's all taken care of.

nice! crewai for the win

You are a kind dude. Thanks for your hard work showing us so much information!

Oh look, another framework that does it "right"...

I don't think we need to set os.environ again after calling load_dotenv()?

is this new autogen nerfed? there are only 3 available model types in my version (0.1.3) -- openai, google, azure

Greetings from Brazil. Two weeks trying all sorts os tutorials until getting across this one here. Fantastic video! Finally made it - all thanks to you! Will now work on the Crew I really want to, using web broweser and URLs scraper tools. Let's see how it goes. Thank you Tyler.

I'm new to AI and "agents," but I recently coded a poem-generating script to learn Python and the related processes. It was a steep learning curve, but I have to say, your explanations are the best I've seen by far. Keep up the good work! I'm 57 years old, and I'm really excited about the potential of AI and what we can achieve with it. As a full-time trader specialising in gold (XAUUSD), I want to use AI to assist me with chart analysis. 😃

how do we go function calling in when using ollama llama 2?

This is a great video thank you, i have been playing with autogen and crewai for a while now and really loving it, quick tip for people i dont think this code is on the github yet and if your lazy like me and dont want to type just take a screen shot of the code in the video send it to gpt and ask it to print it out as text and pow you can easy copy and paste

hey Tyler right now, what do you think is going into the best direction autogen or crewai? im starting a long term project and i value your opinion on which to choose, thanks!!!!!❤

I like Ollama, I like CrewAi, and I enjoy the great content you provide. Thank you 🌹🌹🌹.. Oh I can not understand you people, 121 views and only 14 likes !!! shame on you people .. You all know that the good content in CZcams is truly rare .. We must support the good providers such as Mr. Tyler.. Thanks again Mr. Tyler .. Keep it good 🌹🌹🌹

Nice Taylor.. It's clear to understand

Thank you for sharing this, but i want to see the complete prompt to the LLM when i am using OpenAI models, is there a way i can check that? I understand chat_results gives me a content role and any other information, which i believe is just openAI chat completion output. I want to print the input to the LLM and the output from the llm, ofcourse not the mbeddings but the text. is there a way to expose that infromation in autogen?

Is the liteLLM step still required? And how would you add claude to autogen studio?

i am getting this error "[{'type': 'extra_forbidden', 'loc': ('body', 'functions'), 'msg': 'Extra inputs are not permitted', 'input': [{'name': 'research', 'description': 'research about a given topic, return the research material including reference links', 'parameters': {'type': 'object', 'properties': {'query': {'type': 'string', 'description': 'The topic to be researched about'}},

Great ❤❤❤

i don't why but it's not using tools to fetch information ``` from autogen import AssistantAgent,UserProxyAgent,GroupChat,GroupChatManager,ConversableAgent import os from dotenv import load_dotenv from search_tool import search_internet import json load_dotenv() # config_list=[ # { # "model":"meta.llama3-70b-instruct-v1:0", # "api_type":"open_ai", # "base_url":os.environ['OPENAI_BASE_URL'], # "api_key":os.environ['OPENAI_API_KEY'] # } # ] config_list=[ { "model":"meta/llama3-70b-instruct", "api_type":"open_ai", "base_url":"integrate.api.nvidia.com/v1", "api_key":"" } ] llm_config={ "timeout":300, "cache_seed":42, "config_list":config_list, "temperature":0.6 } def termination_msg(x): return isinstance(x, dict) and "TERMINATE" == str(x.get("content", ""))[-9:].upper() topics_generator = UserProxyAgent( name="BlogTopicGenerator", is_termination_msg=termination_msg, human_input_mode="NEVER", code_execution_config=False, description="An Agent which generates topics for the Blog that it is asked to write", ) examples_adder = AssistantAgent( name="BlogExamplesAdder", is_termination_msg=termination_msg, system_message="You read the json having an article written. Your job is to add more examples in that json. Add a new key Examples and then add atleast 4 examples to it. ", llm_config=llm_config, description="Adding more examples", ) advanced_topic = AssistantAgent( name="AdvancedTopic", is_termination_msg=termination_msg, system_message="You read the json having an article written. Your job is to add a new key which talks about advanced topics related to the article. Give final response in the same JSON format and then reply with TERMINATE", llm_config=llm_config, description="Adding a section on Advanced Topics", ) article_writer = AssistantAgent( name="ArticleWriter1", is_termination_msg=termination_msg, system_message="""You are a technical blog writer, you write an article on the topics provided by BlogTopicGenerator. Each topic should be written with a simple example. Make sure the final response that you give should be in JSON format so that it can be rendered in the UI. It should follow an example structure given below: \{ "Title":"Unraveling Numpy Broadcasting With Clear Examples", "Url":"example.com/blog" "Body":\[ \{ "Content": "In the vibrant world of Python programming, Numpy stands as a fundamental library that empowers its users with extensive mathematical capabilities. Among these wonderful features of Numpy, a little appreciated yet pivotal one is 'Broadcasting'. This article will vividly explain and illustrate this fairly perplexing aspect of Numpy." \}, \{ "Heading": "What is Numpy Broadcasting?", "Content": "Numpy broadcasting is a powerful feature that allows us to perform arithmetic operations on arrays of different shapes. This could be between a scalar and an array, or between arrays of different shapes, under certain conditions." \}, \{ "Content": "The concept behind broadcasting pivots on two rules:", "List": \[ \{ "ListItem": "If the arrays do not have the same rank, then a 1 will be pre-pended to the smaller ranking array’s shape." \} \] \}, \] \}""", llm_config=llm_config, description="technical blog writer who writes article on the list of topics provided", ) def state_transition(last_speaker, groupchat): messages = groupchat.messages if last_speaker is topics_generator: return article_writer elif last_speaker is article_writer: return examples_adder elif last_speaker is examples_adder: if messages[-1]["content"] == "exitcode: 1": return article_writer else: return advanced_topic elif last_speaker == advanced_topic: return None def _reset_agents(): topics_generator.reset() article_writer.reset() examples_adder.reset() advanced_topic.reset() topics_generator.clear_history() article_writer.clear_history() examples_adder.clear_history() advanced_topic.clear_history() def create_article(PROBLEM): _reset_agents() groupchat = GroupChat( agents=[topics_generator, article_writer, examples_adder, advanced_topic], messages=[], max_round=4, speaker_selection_method=state_transition, allow_repeat_speaker=False, enable_clear_history=True ) manager = GroupChatManager(groupchat=groupchat, llm_config=llm_config, max_consecutive_auto_reply=1) topics_generator.initiate_chat( manager, message=PROBLEM, ) article_writer.register_for_llm(name="get_information",description="get the infomation about the relevant blogs")(search_internet) article_writer.register_for_execution(name="get_information")(search_internet) keywords = ["Machine learning", "health"] final_json = [] for i, topic in enumerate(keywords): PROBLEM = f"Write an article similar to a blog on {keywords}" create_article(PROBLEM) cleaned_json_string = topics_generator.last_message()["content"].replace("TERMINATE", "").strip() data = json.loads(cleaned_json_string) final_json.append(data) print(data) final = {} for f in final_json: final = {**final, **f} ```

really loved this video!

@TylerAI : I could really use your help on a group chat project that i am doing for work, its a one page app but some complex task provided to group manager, it does half right but need this working asap for office project to cement Autogen to automate stuff. let me know if you can spend like 30 minutes. thanks.

This doesn’t work! Please stop making bs content

Worst onboarding EVER! Can't even sign up for a new account!? Wanted me to stay on the same ? browser to validate via email and when I tried to do sign up again. Yahhh.... 403 Error! NO Thanks! 5 minutes I cannot ever get back. Good Luck with that!

@Tyler AI : I tried Agentops group chat example you provided and I keep getting this error "🖇 AgentOps: Failed to set up AutoGen logger with AgentOps. Error: start() got an unexpected keyword argument 'logger'" on the very first line and dont get any session ID created. I tried the same lines of AgentOps code in my other projects has the same issue.

My biggest concern, even if autogen, langchain, llama.index(GPT/Microsoft) offer these agents, what is the guarantee, that they won't start charging for these ideas? I'm assuming this is coming down the pipeline soon. I was watching another guy mention that pipelines is similar to agents, and that's how I've decided so far, since the Agents require so much more than just a simple prompt to the model directly. I'll say that langchain, and autogen, I tried and struggled with the configurations and dependancies. I'm slowly learning what works and what doesn't but it's pretty obvious to me that neither of them like windows, and also neither of them like the latest versions of python environments. You would think that someone had created a program that could have all the dependencies automatically integrated and working rather than everyone having to diagnose this issue individually. I'm finding this task the biggest issue to getting these LLM apps to work on my Windows computer without having to setup the linux environment...

This was beautifully performed from your lovely voice to the stings 💛💛💛😩. Chills throughout the entire video !! Hella underrated. PERIOD

Cool

very interesting interview! now with TylerTV we close the circle and the IPO for TylerCorp is next week 🦾🦾🦾

I am not getting the py file at 10:54. the files are not saving for me. i am on mac too.

when to use register_for_llm and register_for_execution ?

is function calling same as adding skills to agents ? if not can you make a video on adding skills to agents in autogen

When I install autogenstudio (beta) I am NOT seeing "Gallery". Is there something during install I am missing?