INDEX

Explanations

phrases related to racks or stacking items

oai_token-act-pair · gpt-3.5-turbo

references to the word "rack" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Austral

-0.74

 wills

-0.63

 Samoa

-0.61

éĸ

-0.60

 Pione

-0.60

 Lover

-0.60

iating

-0.59

 Kashmir

-0.59

 Broadway

-0.58

NSW

-0.58

POSITIVE LOGITS

cloth

1.06

washer

1.03

acks

0.98

 racks

0.97

yard

0.97

ings

0.94

TOP

0.93

mount

0.91

artisan

0.90

 rack

0.90

Activations Density 0.014%

phrases related to racks or stacking items

references to the word "rack" in various contexts

No Comments

No Known Activations