INDEX

Explanations

references to the name "Henson"

oai_token-act-pair · gpt-3.5-turbo

mentions of the individual Jim Henson

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

 base

-0.68

HD

-0.66

Lv

-0.65

 divided

-0.64

 justice

-0.63

cos

-0.61

UV

-0.61

 Geral

-0.60

ised

-0.59

 terminal

-0.59

POSITIVE LOGITS

enson

4.67

ensen

2.32

inson

1.38

enegger

1.15

ens

1.10

eson

1.06

enberg

1.03

ersen

0.99

enstein

0.98

ERSON

0.98

Activations Density 0.006%

references to the name "Henson"

mentions of the individual Jim Henson

No Comments

No Known Activations