INDEX

Explanations

words related to legal or criminal proceedings, such as charges, sentencing, and arrests

oai_token-act-pair · gpt-3.5-turbo

forms of the verb "to be" in relation to legal circumstances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

inav

-0.85

velt

-0.75

acly

-0.74

ankind

-0.73

odes

-0.70

eways

-0.68

 prophecy

-0.68

PsyNetMessage

-0.67

 Echoes

-0.67

Liter

-0.66

POSITIVE LOGITS

 sentenced

1.74

 indicted

1.71

 acquitted

1.68

 jailed

1.65

 arrested

1.62

 arra

1.61

 convicted

1.58

 charged

1.50

 prosecuted

1.44

 fined

1.43

Activations Density 0.190%

words related to legal or criminal proceedings, such as charges, sentencing, and arrests

forms of the verb "to be" in relation to legal circumstances

No Comments

No Known Activations