INDEX

Explanations

the word "her" with high activation

oai_token-act-pair · gpt-3.5-turbo

the word "her" 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 @ 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

 Reconstruction

-0.69

 Governments

-0.61

 cancell

-0.56

TTC

-0.55

 stopp

-0.55

 Enhance

-0.53

 Origins

-0.53

 tilt

-0.53

CVE

-0.53

 Quarterly

-0.52

POSITIVE LOGITS

her

4.74

hers

3.26

HER

2.45

hes

2.08

hest

2.02

hed

1.72

hel

1.68

hen

1.63

heres

1.61

here

1.56

Activations Density 0.014%

the word "her" with high activation

the word "her" in various contexts

No Comments

No Known Activations