Rakesh

None

[Rakesh @_Rakeshhhhhhh]

BAYC floor possible bottom around 24 ETH +-8ETH Error

[Dr Khushboo @khushbookadri]

Ready to win a luxurious soap hamper from https://t.co/TdcVKYzUxE? 🛁 Follow these steps:

1️⃣ FOLLOW THIS ACCOUNT ✅
2️⃣ Retweet This Tweet ✅
3️⃣ Use #Kbathbrewery ✅
4️⃣ Predict VIRAT KOHLI’s score in #INDvSL Finals 🏏

ONE LUCKY PERSON with the correct prediction will be the winner! 🎉 Only one prediction per person.#AsiaCup2023 #AsiaCupFinal

[Rakesh @_Rakeshhhhhhh]

@khushbookadri 54

[Taelin @VictorTaelin]

Does anyone have connections to researchers or major labs studying Long Covid, insomnia, progressively worsening fatigue etc.?

[Robert Nishihara @robertnishihara]

This in-depth case study sheds light on when you can achieve GPT-4 level performance with a fine-tuned 7B parameter model.

Take SQL generation as an example.

Accuracy
🧿 Llama-2-7B: 3%
🧿 GPT-4: 79%
🧿 Llama-2-7B (fine-tuned): 86%

Out of the box, GPT-4 crushes Llama-2 (including the 70B parameter model), but when fine-tuned, Llama-2-7B wins.

That's superior accuracy with a model 2% of the size of GPT-4 (purportedly), therefore orders of magnitude cheaper. 🤯🤯🤯

This is an astounding lift 🙌🙌

But when does this work? 🤔

The math problem solving benchmark tells a different story.

Fine-tuning still gives a meaningful lift (doubling accuracy for the 7B model), but even the fine-tuned 70B model achieves only 62% accuracy (compared with 98% for GPT-4!!!). 🫢

The study goes into much more detail about when you might expect fine-tuning to give a big win, and when it still won't be enough.

https://t.co/7A6QqJxUqF

[Rakesh @_Rakeshhhhhhh]

@screentimes 0x7b97051fD964B30B9C32A8170f8297c5AbE4324B

[Rakesh @_Rakeshhhhhhh]

@IamNomad Which recorded data you use for understanding Market Participant behaviour

[Rakesh @_Rakeshhhhhhh]

@Via_Osmosis @HPOS10iCASH @housemoneyapp @Bazedgames Thankyou

[Rakesh @_Rakeshhhhhhh]

@Via_Osmosis @HPOS10iCASH Can you share $house twitter handle

[Rakesh @_Rakeshhhhhhh]

@thegreatola Gm

[Rakesh @_Rakeshhhhhhh]

@simpelyfe Combinations of datasets not get you

[MOG COIN @mogcoin]

To celebrate the start of the month of MOGUST we will be raffling off this egirl’s Remilio NFT that she sold at the stone cold bottom. Just like/retweet to be entered into the draw - winner chosen in 24 hours

A very special month for $MOG with many more things to come!

[Rakesh @_Rakeshhhhhhh]

@BloomCapital_ Pepe

[Rakesh @_Rakeshhhhhhh]

@thegreatola 36

[Rakesh @_Rakeshhhhhhh]

@thegreatola 5/6 figure

[Rakesh @_Rakeshhhhhhh]

@thegreatola Wagmi

[Vertigo_Warrior @VertigoWarrior]

Family of Bhagwan Shree Ram

1. Maa Sita - Wife

[Desi Thug @desi_thug1]

10 Must Try Delicious Dishes from Vrindavan

1. Mathura ka Peda

[Rakesh @_Rakeshhhhhhh]

@BloomCapital_ 83034.eth

[Tesla Owners Silicon Valley @teslaownersSV]

🐹 @_hamster_coin 🔥

[Taelin @VictorTaelin]

Letting you know that Claude-2-100k has been remarkably better than GPT-4 for the majority of my non-code queries. Try it!