INDEX

Explanations

time-sequenced events or actions mentioned in a sequence

oai_token-act-pair · gpt-3.5-turbo

instances of the word "followed" indicating sequences or lists

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

adin

-0.69

voy

-0.69

chio

-0.64

inese

-0.64

inia

-0.62

utor

-0.61

uz

-0.61

 Ages

-0.61

 privilege

-0.61

agers

-0.61

POSITIVE LOGITS

 closely

0.96

 suit

0.93

suit

0.87

ĸļ

0.80

SHIP

0.72

 logically

0.70

 thereafter

0.70

edient

0.69

 footsteps

0.68

 shortly

0.65

Activations Density 0.036%

time-sequenced events or actions mentioned in a sequence

instances of the word "followed" indicating sequences or lists

No Comments

No Known Activations