INDEX

Explanations

web-related terms with symbols such as '+', '-' or '*'

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly variations of the plus sign and other symbols

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 skelet

-0.61

aneous

-0.60

PM

-0.60

FAR

-0.59

 tsunami

-0.58

ľ

-0.56

NCT

-0.56

oÄŁ

-0.55

Reply

-0.55

 planted

-0.55

POSITIVE LOGITS

oice

0.86

olor

0.80

 Flavoring

0.78

ibrary

0.77

olkien

0.75

ihilation

0.75

vernment

0.75

redd

0.74

clair

0.74

iverpool

0.73

Activations Density 0.030%

web-related terms with symbols such as '+', '-' or '*'

punctuation marks, particularly variations of the plus sign and other symbols

No Comments

No Known Activations