INDEX

Explanations

phrases related to critical or impactful findings

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

negative aspects or criticisms related to social and political issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

tyr

-0.63

ADRA

-0.58

BRE

-0.56

jong

-0.55

artney

-0.54

ailability

-0.54

senal

-0.53

IDA

-0.53

TIT

-0.53

cill

-0.53

POSITIVE LOGITS

is

1.08

1.03

?:

0.97

isn

0.95

 lies

0.95

 arises

0.94

 here

0.90

 seems

0.87

was

0.85

?ãĢį

0.85

Activations Density 0.150%

phrases related to critical or impactful findings

negative aspects or criticisms related to social and political issues

No Comments

No Known Activations