INDEX

Explanations

terms related to legal accusations or charges, specifically the word "indictment."

oai_token-act-pair · gpt-3.5-turbo

references to legal indictments and related legal terminology

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

birth

-0.87

 Kers

-0.72

 Rapt

-0.71

atan

-0.71

pper

-0.65

 lear

-0.65

istic

-0.64

ooked

-0.64

 Orth

-0.63

erved

-0.63

POSITIVE LOGITS

 indict

1.15

 indictment

1.12

 indicted

1.00

eering

0.99

 summ

0.85

 warrant

0.80

 warrants

0.80

 unse

0.79

 affidav

0.78

 affidavit

0.77

Activations Density 0.019%

terms related to legal accusations or charges, specifically the word "indictment."

references to legal indictments and related legal terminology

No Comments

No Known Activations