INDEX

Explanations

phrases related to legal rules and provisions

oai_token-act-pair · gpt-3.5-turbo

bracketed or parenthetical notations used in legal or formal texts

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

 stare

-0.66

miah

-0.64

 Meridian

-0.62

 gallery

-0.62

 Romero

-0.60

 haunt

-0.60

 Tall

-0.60

 dust

-0.60

 chant

-0.59

 Dare

-0.59

POSITIVE LOGITS

cases

0.98

case

0.90

formerly

0.86

bands

0.85

knit

0.83

*)

0.81

except

0.80

served

0.79

impl

0.79

Solution

0.79

Activations Density 0.019%

phrases related to legal rules and provisions

bracketed or parenthetical notations used in legal or formal texts

No Comments

No Known Activations