INDEX

Explanations

numerical references to the quantity "one."

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "one."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ooks

-0.86

folk

-0.79

lov

-0.79

its

-0.78

ãĥ¼

-0.76

acements

-0.76

late

-0.75

æĸ¹

-0.73

aqu

-0.72

istas

-0.71

POSITIVE LOGITS

 hundred

1.13

 dimensional

1.08

 Hundred

0.94

 sided

0.93

 thousand

0.88

 Piece

0.84

 month

0.84

 hour

0.82

 million

0.81

 piece

0.81

Activations Density 0.066%

numerical references to the quantity "one."

mentions of the word "one."

No Comments

No Known Activations