INDEX

Explanations

phrases related to activities or events happening at a specific time or while something else is occurring

oai_token-act-pair · gpt-3.5-turbo

references to the subject in a narrative or discussion format

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

 PLUS

-0.82

 Christy

-0.74

 GOODMAN

-0.69

 Clarks

-0.64

 Regist

-0.61

 doorstep

-0.61

 Zionism

-0.61

 Santana

-0.61

 Rout

-0.60

Planet

-0.59

POSITIVE LOGITS

phthal

0.68

obi

0.68

nown

0.66

Others

0.66

iliary

0.65

 sleeps

0.65

olean

0.63

empl

0.63

 ordinarily

0.63

 pretended

0.62

Activations Density 0.211%

phrases related to activities or events happening at a specific time or while something else is occurring

references to the subject in a narrative or discussion format

No Comments

No Known Activations