INDEX

Explanations

phrases indicating a preference for one option over another

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize a preference for alternatives or comparisons

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

iard

-0.64

mberg

-0.63

 Coffee

-0.60

andal

-0.57

endale

-0.57

esi

-0.57

adium

-0.57

 DRAG

-0.56

amba

-0.55

enos

-0.55

POSITIVE LOGITS

 than

0.96

than

0.82

iving

0.68

acent

0.66

rame

0.65

 preferring

0.65

 trivial

0.63

othes

0.62

 frivolous

0.62

usions

0.61

Activations Density 0.022%

phrases indicating a preference for one option over another

phrases that emphasize a preference for alternatives or comparisons

No Comments

No Known Activations