INDEX

Explanations

quantitative terms of comparison indicating significance or degree

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate equivalence or comparisons

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

 craw

-0.65

 sled

-0.62

odder

-0.62

 Mansion

-0.62

 fixation

-0.60

away

-0.60

 itch

-0.58

 recess

-0.58

Tid

-0.58

inelli

-0.57

POSITIVE LOGITS

awei

0.90

entially

0.89

 amounts

0.86

ume

0.83

encies

0.83

umes

0.82

externalActionCode

0.80

ngth

0.80

uates

0.75

Pres

0.73

Activations Density 0.038%

quantitative terms of comparison indicating significance or degree

phrases that indicate equivalence or comparisons

No Comments

No Known Activations