INDEX

Explanations

phrases related to intentions and reasons

oai_token-act-pair · gpt-3.5-turbo

phrases indicating certainty, intent, and lack of evidence

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

ahime

-0.79

pherd

-0.65

ortment

-0.62

tsky

-0.62

iets

-0.61

igs

-0.61

visor

-0.60

types

-0.60

acerb

-0.60

ork

-0.60

POSITIVE LOGITS

 whatsoever

1.84

nor

1.15

 anymore

0.95

 anywhere

0.85

 hesitation

0.80

 anybody

0.77

 except

0.75

 slightest

0.75

 EVER

0.74

 necessarily

0.74

Activations Density 0.169%

phrases related to intentions and reasons

phrases indicating certainty, intent, and lack of evidence

No Comments

No Known Activations