INDEX

Explanations

news article headlines with the word "READ" in them

oai_token-act-pair · gpt-3.5-turbo

instances of the word "READ" 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

 Ghana

-0.75

oteric

-0.72

 retirement

-0.67

lux

-0.67

 Bret

-0.62

ylum

-0.61

 legitimate

-0.61

ugi

-0.60

ofi

-0.59

ascal

-0.59

POSITIVE LOGITS

READ

1.08

 READ

0.98

 ALSO

0.91

 aloud

0.89

SEE

0.89

MORE

0.88

BOOK

0.86

LESS

0.86

NESS

0.82

htaking

0.82

Activations Density 0.006%

news article headlines with the word "READ" in them

instances of the word "READ" in various contexts

No Comments

No Known Activations