INDEX

Explanations

phrases or words related to evaluations or judgments

oai_token-act-pair · gpt-3.5-turbo

terms related to moral and ethical qualifications

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

berman

-0.78

 Denmark

-0.73

shire

-0.73

LAND

-0.69

avorite

-0.65

 Webster

-0.65

hire

-0.64

 Sutherland

-0.63

 Danish

-0.62

 Cumber

-0.62

POSITIVE LOGITS

itatively

1.15

ifications

1.08

ific

0.98

ifying

0.96

ãĥ£

0.93

ifiable

0.93

ifiers

0.92

ifi

0.91

ified

0.90

ities

0.89

Activations Density 0.029%

phrases or words related to evaluations or judgments

terms related to moral and ethical qualifications

No Comments

No Known Activations