INDEX

Explanations

words related to the suffix '-ot'

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "not."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 fixme

-0.66

 silence

-0.63

surface

-0.62

TextColor

-0.61

 reigning

-0.61

 discredit

-0.60

 toast

-0.59

 sauces

-0.58

 Keen

-0.57

 readiness

-0.57

POSITIVE LOGITS

ot

1.14

arget

1.05

tery

1.03

oad

1.01

oot

1.00

inyl

0.98

anu

0.97

OGR

0.95

ropy

0.95

aylor

0.94

Activations Density 0.013%

words related to the suffix '-ot'

occurrences of the word "not."

No Comments

No Known Activations