INDEX

Explanations

phrases related to the continuation or reading of a main story, potentially for newsletters or articles

oai_token-act-pair · gpt-3.5-turbo

instances of the word "story."

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

inking

-0.85

 stocking

-0.83

ipher

-0.82

ocol

-0.80

omin

-0.78

itions

-0.78

omination

-0.78

iling

-0.74

istered

-0.73

othing

-0.73

POSITIVE LOGITS

 Stories

0.85

 Surviv

0.81

LIN

0.79

 story

0.78

 arcs

0.78

 GOODMAN

0.77

\\\\\\\\

0.77

story

0.76

 stories

0.76

arc

0.76

Activations Density 0.035%

phrases related to the continuation or reading of a main story, potentially for newsletters or articles

instances of the word "story."

No Comments

No Known Activations