INDEX

Explanations

phrases related to sending information or recommendations to friends or colleagues

oai_token-act-pair · gpt-3.5-turbo

articles and possessive pronouns

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

tery

-0.77

packages

-0.68

/-

-0.62

quickShipAvailable

-0.62

 Theft

-0.59

.–

-0.59

Split

-0.58

 clues

-0.58

 indications

-0.57

 chunks

-0.56

POSITIVE LOGITS

venge

1.28

 detriment

1.11

 extent

1.03

rouse

1.02

 fullest

0.87

 tune

0.85

merce

0.82

 liking

0.79

 mailbox

0.77

 clipboard

0.76

Activations Density 0.260%

phrases related to sending information or recommendations to friends or colleagues

articles and possessive pronouns

No Comments

No Known Activations