INDEX

Explanations

words related to behind-the-scenes activities or occurrences

oai_token-act-pair · gpt-3.5-turbo

phrases related to being "off the grid" or disconnected from mainstream trends and media

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Seeking

-0.73

Tik

-0.65

 fade

-0.63

 thanking

-0.61

 abruptly

-0.60

 pending

-0.59

 Rica

-0.59

 terminated

-0.58

 dwind

-0.56

 sighting

-0.56

POSITIVE LOGITS

prem

1.06

same

1.05

middle

0.96

original

0.95

surface

0.95

concept

0.94

distance

0.93

package

0.92

best

0.90

bone

0.88

Activations Density 0.011%

words related to behind-the-scenes activities or occurrences

phrases related to being "off the grid" or disconnected from mainstream trends and media

No Comments

No Known Activations