INDEX

Explanations

words related to television programs or broadcasts

oai_token-act-pair · gpt-3.5-turbo

references to time or events associated with the concept of "lifetime."

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

FU

-0.77

 chim

-0.62

bug

-0.61

 concerned

-0.58

 naked

-0.57

 divisions

-0.57

 sign

-0.55

 annex

-0.55

imp

-0.55

 robots

-0.55

POSITIVE LOGITS

etime

4.83

etimes

2.18

elong

1.27

eworld

1.20

estate

1.19

elight

1.16

ename

1.15

estyle

1.13

ftime

1.12

 Lifetime

1.12

Activations Density 0.016%

words related to television programs or broadcasts

references to time or events associated with the concept of "lifetime."

No Comments

No Known Activations