INDEX

Explanations

actions or processes described in the form of "to [verb]"

oai_token-act-pair · gpt-3.5-turbo

phrases related to understanding and making things

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ´ãĥ³

-0.62

keyes

-0.56

selling

-0.56

wed

-0.56

tch

-0.56

 Travels

-0.55

uploads

-0.55

Apr

-0.55

judicial

-0.55

 Dresden

-0.55

POSITIVE LOGITS

uate

0.78

 properly

0.75

 truly

0.71

 this

0.70

 further

0.70

 your

0.69

irtual

0.65

ISE

0.64

 fullest

0.64

eeper

0.63

Activations Density 0.119%

actions or processes described in the form of "to [verb]"

phrases related to understanding and making things

No Comments

No Known Activations