INDEX

Explanations

phrases related to things that are unidentified or not accounted for

oai_token-act-pair · gpt-3.5-turbo

terms related to lack of recognition or accountability

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

cision

-0.63

arb

-0.62

 Kills

-0.62

 Rite

-0.61

 Pigs

-0.61

strike

-0.61

 Farmers

-0.60

EFF

-0.60

ulton

-0.60

 Bulls

-0.59

POSITIVE LOGITS

 unaccount

1.42

ilater

0.94

aband

0.85

ancies

0.84

atile

0.83

untarily

0.82

 unrecogn

0.80

esy

0.79

ileaks

0.77

 incorpor

0.75

Activations Density 0.003%

phrases related to things that are unidentified or not accounted for

terms related to lack of recognition or accountability

No Comments

No Known Activations