INDEX

Explanations

mentions of criminal charges or crimes

oai_token-act-pair · gpt-3.5-turbo

instances of the word "charged" in the context of legal accusations

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

 tasted

-0.63

quickShipAvailable

-0.62

echo

-0.61

Ħ¢

-0.60

strings

-0.59

Zone

-0.59

vation

-0.58

 tastes

-0.57

 adapted

-0.57

 fades

-0.57

POSITIVE LOGITS

 criminally

1.09

heet

0.97

llah

0.92

 with

0.82

 indict

0.77

 indictment

0.76

 arra

0.74

 alleging

0.72

 felony

0.72

 indicted

0.71

Activations Density 0.033%

mentions of criminal charges or crimes

instances of the word "charged" in the context of legal accusations

No Comments

No Known Activations