Dr. Daniel Bender pfp

Dr. Daniel Bender

@aidful

5 Following
3 Followers
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Stay ahead in your field as AI transforms industries. With technology evolving rapidly, you must keep pace with the latest trends to understand AI's impact on your sector. If you do, you will be one of the winners! 🦾πŸ’ͺ
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Google has YouTube, the best source for future training, with tons of new daily content. I am confident that they will manage to copy the models from others and apply it to their data. What do you think?
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
AI powerhouses Mustafa Suleyman and Karen Simonyan, co-founders of Inflection AI, leave their $1.3 billion startup to lead Microsoft's new AI division. Do you think Sam Altman will be happy about this?
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ”’πŸ’¬ Privately chat with multiple open-source models at once The new LM Studio version 0.2.17 makes this possible and the only limit is your GPU memory! You can also use the same model in more than one quantization or with different system prompts and judge which works best.
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ”πŸ–Ό New transparency feature in leonardo.ai β€’ Now you get a true native transparent PNG generation. β€’ Not just background removal, it offers partial transparency and clean edges.
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ€–πŸ†“ Use ALL Claude 3 variants for FREE! Claude 3 Sonnet is free. Haiku (fastest) and Opus (best) are available through paid API or Pro subscription from @AnthropicAI BUT, you can access all Claude 3 variants for free at the Chatbot Arena: arena.lmsys.org/
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
βœοΈπŸ’‘ DeepSeek-VL 7B: A free GPT-4 Vision Competitor The performance to recognize my handwriting is slightly better than GPT-4. Impressive! The DeepSeek Model License aims to prevent misuse of its AI capabilities. Test it on: huggingface .co/deepseek-ai/deepseek-vl-7b-chat
0 reply
0 recast
2 reactions
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ“ˆπŸ†“ Use Claude 3 Opus for FREE: Claude 3 Opus outperforms GPT-4 in many scenarios, but it's only available with a paid subscription. Access it for free at "arena.lmsys .org" and follow the steps shown in the image. πŸ‘‡
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ”ŽπŸ˜― AI integration will be the game changer! The you.com search-engine makes it easy to extract info from YouTube videos (and other docs) with their Genius mode. Compare the result from asking Genius a question of a nearly 3-hour long video with the transcript. Pure gold! πŸ‘‡
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Thanks for sharing. Would love a bit more context. I did not know it and looked it up:
1 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
I did one and a half "100 Days Of AI" challenges, and now I am thinking about creating my own short challenge to solve a real problem. Tell me if there is anything you are eager to learn in a guided 7 or 10-day challenge. πŸ‘‡ https://x.com/AidfulAI/status/1767096304008613947?s=20
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ—£πŸ“ Prompt Whisper, OpenAI's audio2txt model, to get better results. You can improve the already good audio transcriptions from Whisper by prompting the model. The maybe best feature is to add technical terms like GPT-4 or Gemini. Source: https://platform.openai.com/docs/guides/speech-to-text/prompting
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸ†“ Gemini 1.0 Pro API usage is free of charge But you pay with your data, as your chats are used to train future models. Nevertheless, good to test the model in an application like https://www.typingmind.com, which made me aware of this in their newsletter.
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
πŸŽ₯πŸ’‘ Many insights are hidden in YouTube content. Looking forward to getting the info w/o watching a long video? Chatting with video transcripts is just the beginning. LLMs will soon revolutionize how they extract both text and visuals, taking results to the next level. Easiest way today: https://glasp.co
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Soon you will be able to extract the information you are looking for by chatting with a video. To some extend you already are by using the transcript as input to an LLM. Using the visual information on top will make results even better in the future.
0 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
You are right, but hard to avoid collisions like the one you named if two fields are combined. Often the context will solve the confusion.
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Do you want to understand the concept behind latent space or do you have another intention? To understand it, I think the Wiki article is quite good to get started: https://en.m.wikipedia.org/wiki/Latent_space
1 reply
0 recast
1 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
According to Wikipedia, Mario first appeared in Donkey Kong 1981. After Donkey Kong, Mario starred in Mario Bros 1983. In 1985 NES sequel, Super Mario Bros., began the successful Super Mario series.
0 reply
0 recast
0 reaction
Dr. Daniel Bender pfp
Dr. Daniel Bender
@aidful
Also Google, Mistral and Meta have models which they did not release as open-source. From OpenAI we got, besides other releases, Whisper (speech2txt) as open-source. Nevertheless, I would love more open models and especially an LLM from them. Like from Meta and now, with Gemma, also Google.
0 reply
0 recast
0 reaction