INDEX

Explanations

phrases related to criminal charges or accusations

oai_token-act-pair · gpt-3.5-turbo

frequently occurring articles and pronouns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ilaterally

-0.77

edIn

-0.73

æµ

-0.66

emale

-0.65

ãĥ¼ãĥĨ

-0.63

itud

-0.63

Interstitial

-0.63

ãĥĩãĤ£

-0.62

alore

-0.62

pheus

-0.62

POSITIVE LOGITS

 Coverage

0.62

ISS

0.55

esson

0.54

ANS

0.53

osures

0.53

 Â®

0.52

CRIPTION

0.51

 Violent

0.50

lement

0.50

REL

0.49

Activations Density 0.996%

phrases related to criminal charges or accusations

frequently occurring articles and pronouns

No Comments

No Known Activations