INDEX

Explanations

movie titles, specifically focusing on the presence of the word "starring"

oai_token-act-pair · gpt-3.5-turbo

references to actors and their roles in films

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

alde

-0.70

EAR

-0.68

upon

-0.63

 individual

-0.62

oat

-0.62

IB

-0.61

 instr

-0.61

abis

-0.60

XT

-0.60

 individuals

-0.59

POSITIVE LOGITS

 starring

3.74

 starred

1.65

 featuring

1.52

 cameo

1.13

 stars

1.09

 Featuring

1.09

 showcasing

1.07

 portraying

1.05

 depicting

1.04

stars

1.04

Activations Density 0.014%

movie titles, specifically focusing on the presence of the word "starring"

references to actors and their roles in films

No Comments

No Known Activations