INDEX

Explanations

quotes or phrases indicated by the word "quote"

oai_token-act-pair · gpt-3.5-turbo

references to quotes

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

 Measures

-0.74

ead

-0.72

hea

-0.65

ateurs

-0.65

icult

-0.62

ollen

-0.61

dom

-0.61

aval

-0.61

impro

-0.60

allows

-0.59

POSITIVE LOGITS

 quote

4.11

 quotes

2.80

 quotation

2.60

 Quote

2.38

quote

2.31

 quotations

2.19

 quoting

2.04

Quote

1.91

 quoted

1.86

 paraph

1.81

Activations Density 0.017%

quotes or phrases indicated by the word "quote"

references to quotes

No Comments

No Known Activations