INDEX

Explanations

instances where someone is pointing something out or drawing attention to something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pointed" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ESE

-0.69

ccording

-0.68

DAQ

-0.65

avement

-0.65

uthor

-0.62

apon

-0.62

 Alive

-0.61

aden

-0.61

anny

-0.61

MET

-0.60

POSITIVE LOGITS

 pointed

1.14

 pointing

1.03

 deduction

0.92

 deductions

0.84

 points

0.82

TextColor

0.78

points

0.76

iasis

0.73

point

0.72

dag

0.68

Activations Density 0.018%

instances where someone is pointing something out or drawing attention to something

instances of the word "pointed" in various contexts

No Comments

No Known Activations