INDEX

Explanations

questions posed to the reader

oai_token-act-pair · gpt-3.5-turbo

rhetorical questions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

essential

-0.51

©¶æ

-0.47

athered

-0.45

iazep

-0.44

ulk

-0.42

 temporary

-0.42

 highly

-0.42

umbn

-0.41

 Berm

-0.41

POSITIVE LOGITS

2.87

)?

2.63

'?

2.44

?!

2.41

?:

2.39

?).

2.37

?ãĢį

2.30

!?

2.30

?"

2.28

?'

2.28

Activations Density 0.088%

questions posed to the reader

rhetorical questions

No Comments

No Known Activations