Show HN: less than 650 LOC trainable GPT only using NumPy
1–10 of 19 posts
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#2I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture.
For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#3I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture. For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
The most interesting thing in this whole saga is that decoder only models (aka causal transformers like GPT) are as effective as they are.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#4I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture. For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
FWIW: the GitHub project description says “GPT-like”. It’s the title here that dropped the “like”.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#5I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture. For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
One small difference is that the GPT architecture is just the decoder stack of the original transformer as opposed to the full encoder decoder stack in the original.
I agree the branding play on GPTs in general is pretty smart and strong from OpenAI though.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#6I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture. For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
Regarding regulatory capture, I listened to an interview with Lena Khan, the current head of the FTC, and this exact thing came up as something regulators are worried about. I think regulators are aware of the danger of letting industry insiders regulate their own industry, so I'm hopeful for some sensible regulations that help promote rather than harm competition. The FTC also exists to prevent monopoly.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#7I think people are forgetting that transformer architectures are a wider field from GPT and predate GPT3 by 3+ years. Referring to transformer architectures using a branded commercial nomer (GPT) is just going to help cement OpenAI’s brand exposure and soon regulatory capture. For comparison this would be like referring to convonets as Inception architectures back during the CV boom (or VGGnets before that)
One small difference is that the GPT architecture is just the decoder stack of the original transformer as opposed to the full encoder decoder stack in the original. I agree the branding play on GPTs in general is pretty smart and strong from OpenAI though.
Honestly i feel like the fact that everyone is just calling LLM's GPT at this point doesn't really help OpenAI, ChatGPT would, but the fact is that unlike "googling" something became synonymous for searching on the internet, GPT != OpenAI-ing something, GPT just became what people call LLM's it seems like lately, the fact the term isn't the name of the company or the full name "chatgpt-ing" sort of breaks that hold i feel like.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#8I wonder how easy it would be to port this library from numpy to cupy.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#9Nice! The README mentions `LayerNorm` is implemented here, but while it's in the equivalence tests with PyTorch, I don't see it in the implementation.
Re: Show HN: less than 650 LOC trainable GPT only using NumPy
#10Nice! The README mentions `LayerNorm` is implemented here, but while it's in the equivalence tests with PyTorch, I don't see it in the implementation.
It's part of the TensorLi definition where all the magic happens.