INDEX

Explanations

words related to the direction of events or situations

oai_token-act-pair · gpt-3.5-turbo

references to the status or condition of things

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

iciency

-0.72

essor

-0.65

tein

-0.61

 disav

-0.59

asking

-0.59

¿½

-0.59

ritical

-0.59

 sole

-0.58

 overwrite

-0.57

ividual

-0.57

POSITIVE LOGITS

 downhill

0.77

for

0.75

 between

0.73

 backstage

0.71

 unfold

0.68

roy

0.65

huh

0.64

 diplom

0.63

Things

0.63

FOR

0.62

Activations Density 0.305%

words related to the direction of events or situations

references to the status or condition of things

No Comments

No Known Activations