INDEX

Explanations

names or terms related to rankings

oai_token-act-pair · gpt-3.5-turbo

rankings and scores related to performance

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

flo

-0.89

ãĤ§

-0.88

 pron

-0.86

 fluorescent

-0.77

phas

-0.76

 gluc

-0.76

 skate

-0.75

enza

-0.72

 clut

-0.72

 grun

-0.70

POSITIVE LOGITS

 Rank

1.18

Rank

1.12

 Flavoring

0.91

 Categories

0.89

 Sites

0.87

 Thumbnails

0.82

 Site

0.82

 Rankings

0.82

Constructed

0.81

 Finish

0.80

Activations Density 0.013%

names or terms related to rankings

rankings and scores related to performance

No Comments

No Known Activations