MattVidPro AI
AI News! Timestamps coming soon!
▼ Link(s) From Today’s Video:
✩ Peak into gpt fine tuning: https://twitter.com/OfficialLoganK/status/1704181284036300970
✩ Open AI training bumped to 2022 even though it’s denied: https://twitter.com/sociaIrate/status/1703972889635271046
✩ Reports that Open AI’s new GPT 3.5 instruct is very smart: https://twitter.com/generatorman_ai/status/1703959330448658942
✩ New Fine tuning model: https://platform.openai.com/playground
✩ Open AI hampers chatgpt on ios and android: https://twitter.com/NickADobos/status/1704165692424282175
✩ DALL E 3: https://openai.com/dall-e-3
✩ DALL E 3 Sneak Peak: https://www.youtube.com/watch?v=koR1_JBe2j0&t=729s&ab_channel=MattVidProAI
✩ Eleven labs projects: https://twitter.com/elevenlabsio/status/1704244556672213493?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Etweet
✩ Biggest google Bard update yet: https://twitter.com/DataChaz/status/1704123675036029151
✩ Bard: https://bard.google.com/chat
✩ Google gives people gemini access: https://www.therundown.ai/p/googles-chatgpt-slayer-inches-closer
► MattVidPro Website: https://MattVidPro.com
► MattVidPro Discord: https://discord.gg/bQgcbjs2Sg
► Follow Me on Twitter: https://twitter.com/MattVidPro
————————————————-
▼ Extra Links of Interest:
✩ AI LINKS MASTER LIST: https://www.futurepedia.io/
✩ Lemon Gallery: https://MattVidPro.com/Gallery
✩ General AI Playlist: https://www.youtube.com/playlist?list=PLrfI66qWYbW3acrBQ4qltDBsjxaoGSl3I
✩ AI I use to edit videos: https://www.descript.com/?lmref=nA4fDg
✩ Second Channel: https://www.youtube.com/channel/UC06GdmaEdCdCFwR3NvszloQ
————————————————-
Thanks for watching Matt Video Productions! I make all sorts of videos here on Youtube! Technology, Tutorials, and Reviews! Enjoy Your stay here, and subscribe!
All Suggestions, Thoughts And Comments Are Greatly Appreciated… Because I Actually Read Them.
————————————————-
► Business Contact: MattVidProSecond@gmail.com
Thank you for watching! Timestamps coming soon (ran outta time today) If any of you awesome community members provide some I will ping!
I just asked it for the last 500 characters on Google Chrome on desktop as well, and it read me the information from my custom training set
I asked Bard to pull up email thread about fish. It completely hallucinated a discussion between me and my friend about the benefits of Salmon. That never happened.
The raw models are probably what made experts be afraid of AI
what would happen if you gave gpt4-code interpreter the ability to send requests to gpt4-instruct? probably wouldn't hallucinate as much
I enjoyed your update on bard and chatgpt. Overall great content.
The chatgpt part was way too long
You cannot fine-tune in the true sense. You can't add data, unfortunately… you can only indicate the manner in which you'd like it to respond when prompted.
Just call it "Ex-Twitter"…
Dall-e 3 is a step up from MJ in terms of prompt understanding. Seems about on par with SDXL's text writing ability, and I'm looking forward to comparing them in regards to understanding.
However, in terms of quality MJ, and SDXL, are still on top. Dall-e 3 is not bad by any means, though- it's quite competent.
06:45 INSTRUCT MODELS as holy grail over CHAT MODELS – explained here.
15:00~ ELEVENLABS incredible any voice and multi-lingual STORY PRODUCTION PROJECT TROPE is READY FOR PRIME TIME AUTHOR-DIRECTOR-PRODUCERS.
I asked it for it's initial prompt and it told me "I don't have the ability to access, share, or print specific characters or portions of my prompt, including the last 500 characters."
That didn't take long to block.
GPT3.5 via chrome browser on Android.
Open AI should collaborate with Eleven Labs, so we can hear and talk to ChatGPT in any voice and in any language.
10h 1,1k lies 213kmsybs
notice the knowledge cut off is 2022 instead of 2021 at 10:07 so apparantly did they update then as 2022 doesnt seem to equte into the 2021 response I use to get
We need more info on the best conversational LLM – HeyPi which seem to be extremely up to date. I have even received correct information about how to dock in Starfield. That's pretty recent!
🎯 Key Takeaways for quick navigation:
00:00 🤖 OpenAI News Roundup
05:29 🎨 DALL-E 3 Announced
14:26 🎙️ ElevenLabs Projects Audio Editor
18:10 🤖 Google Bard Extensions
22:10 🤖 Google's Gemini AI
Bilbo reading the silmarillion lol.
nice my partner just got access to the fine tuning ai and his excited
Another great roundup, and that elevenlabs demo is so exciting.
💯 percent down for google renaiming to Barf. I would use it then.
🎉🎉
DALLE3 looks soo good, definitely better than Midjourney V5. I am excited for the next weeks because Midjourney devs are saying that V6 will be a big leap too.
I think if Bard had been a text to music AI it would have been quite a good name but as a chat AI it's terrible.
are you AI yet, I cant tell
so to use it we need to fist pay!
Agreed bard is such a weird choice of name , although the name Gemini sounds wonderful possible strategic approach behind this, image if bard was like the common name it just feels like a pup name/ game character 😂
waiting for big guns
Bard still not available in Canada.
😡
Is there an LLM that can generate accurate technical drawings, diagrams, other visualizations etc. That is what I want and need.
With respect to Gemini: Show me, don't tell me. Bard has been a big disappointment, therefore not expecting much from Gemini. I hope I'm very wrong.
Btw Microsoft announced bing image creator uses dall-e 3
i have seen one day i ask chatgpt something and it takes like3 or 4 prompts to get code i want. then a week later that same piece of code gets given to me on first try.
greetings from spain and thnx for activate subtitles for no english speakers, great channel!
Can I pluck your eyebrows?
The vocal fry on that bard video makes me want to bardf (barf)
Um, I disagree with the gp4 is dumber is just the ios android users…. I use GPT4 for coding assistance and it just … plainly does dumb things.
I'm sure the lobotomization as everyone calls it, greatly hampers its ability to think and generate useful text as a response.
When some one figures out how to find deleted personal information, photos, videos THAT will be evolutionary
Stop doing that funky yeeyee voice bro. just use your voice, you dont have to pretend anymore
Imagine if AI voice stuff gets good enough to have English dubbed animes be voiced by the Japanese voice actors
13:37 that looks like claymation rather than 2d, but oh well.