INDEX

Explanations

names of people, potentially associated with entertainment or news media

oai_token-act-pair · gpt-3.5-turbo

proper nouns, specifically names of individuals or characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

aterasu

-0.67

_-

-0.66

;;;;;;;;;;;;

-0.59

 Grayson

-0.58

 wipes

-0.58

âĶĢâĶĢ

-0.57

PsyNetMessage

-0.55

 ---------

-0.53

 defeats

-0.53

 intersection

-0.52

POSITIVE LOGITS

anski

0.78

cz

0.75

animous

0.71

éŃĶ

0.70

ä»£

0.68

ansky

0.65

ovi

0.64

ski

0.64

ãĤ±

0.64

arella

0.63

Activations Density 0.125%

names of people, potentially associated with entertainment or news media

proper nouns, specifically names of individuals or characters

No Comments

No Known Activations