Content pfp
Content
@
0 reply
0 recast
0 reaction

Gajesh pfp
Gajesh
@gaj
$50 to anyone who can help me train Llama 3.2 Instruct Model and help play around with it 2 days @bountybot
6 replies
8 recasts
17 reactions

Daniel - Bountycaster pfp
Daniel - Bountycaster
@pirosb3
What do you mean by train? are you referring to a fine-tuning process? what is the end goal you would like to see?
1 reply
0 recast
0 reaction

Gajesh pfp
Gajesh
@gaj
yep. fine-tuning. i basically want to train it all my tweets
1 reply
0 recast
0 reaction

Daniel - Bountycaster pfp
Daniel - Bountycaster
@pirosb3
What would the final product look like?
1 reply
0 recast
0 reaction

Gajesh pfp
Gajesh
@gaj
Just a Twitter bot that would be a reply guy and tweet like me
1 reply
0 recast
0 reaction

Daniel - Bountycaster pfp
Daniel - Bountycaster
@pirosb3
Ah! that makes sense. Try doing it without training first. For your bot to be a reply guy, you should train it with many tuples of [request, reply]. Based on the prompt below, and applying a 20% discount, you could pass ~8k tweets in 1 single prompt. So that would be ~4k tuples of [request, reply]. https://chatgpt.com/share/671f0338-54f0-800f-a33c-6eb53492f59e
1 reply
0 recast
0 reaction

Gajesh pfp
Gajesh
@gaj
yep. but that’s just the goal of the product. i wanna learn more about the journey of actually finetuning these models
2 replies
0 recast
1 reaction

Daniel - Bountycaster pfp
Daniel - Bountycaster
@pirosb3
Ah! Love it, have you looked at the fine tuning system available in OpenAI developer panel?
1 reply
0 recast
0 reaction

Gajesh pfp
Gajesh
@gaj
yep but I wanted to try training open source LLMs like Llama3
0 reply
0 recast
1 reaction