INDEX

Explanations

words related to rejection or being uninterested

oai_token-act-pair · gpt-3.5-turbo

phrases indicating opposition or dissent

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

raq

-0.79

soType

-0.76

ased

-0.68

è¦ļéĨĴ

-0.68

forcement

-0.67

CVE

-0.64

racuse

-0.63

aptic

-0.63

azard

-0.63

=>

-0.61

POSITIVE LOGITS

 necessarily

0.95

shy

0.85

 anymore

0.84

 earthly

0.81

 conventional

0.80

 whatsoever

0.80

any

0.80

 anything

0.80

 altogether

0.75

 ever

0.75

Activations Density 0.564%

words related to rejection or being uninterested

phrases indicating opposition or dissent

No Comments

No Known Activations