INDEX

Explanations

parts of programming or technical language, specifically related to variable names and functions

oai_token-act-pair · gpt-3.5-turbo

references to programming concepts, specifically about reserved keywords and variables

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Cadillac

-0.55

Slot

-0.53

 Hutchinson

-0.51

 Vacc

-0.50

 marrow

-0.49

 Entered

-0.48

 Murdoch

-0.46

illac

-0.46

 Hogan

-0.45

Gut

-0.45

POSITIVE LOGITS

heres

0.63

 uncond

0.58

SourceFile

0.58

 neighb

0.57

rex

0.57

-->

0.56

 counterpart

0.55

 PLUS

0.55

HIS

0.55

him

0.55

Activations Density 1.911%

parts of programming or technical language, specifically related to variable names and functions

references to programming concepts, specifically about reserved keywords and variables

No Comments

No Known Activations