INDEX

Explanations

words related to negativity or criticism

oai_token-act-pair · gpt-3.5-turbo

the presence of the substring "pl" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

âĸ¬

-0.74

HAHA

-0.71

shapeshifter

-0.70

QUI

-0.70

DEM

-0.68

é¾įåĸļå£«

-0.67

âĸ¬âĸ¬

-0.66

gerald

-0.65

////////////////////////////////

-0.64

spin

-0.63

POSITIVE LOGITS

asma

1.30

acement

1.18

atinum

1.18

icably

1.12

enty

1.10

astic

1.08

anted

1.02

atter

1.01

atoon

0.99

ague

0.98

Activations Density 0.009%

words related to negativity or criticism

the presence of the substring "pl" in various contexts

No Comments

No Known Activations