INDEX

Explanations

words or phrases related to legal actions and decisions

oai_token-act-pair · gpt-3.5-turbo

references to legal decisions or actions that affirm or maintain laws and rights

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

NetMessage

-0.87

goo

-0.81

apter

-0.78

ohydrate

-0.71

pox

-0.70

plet

-0.67

ascus

-0.64

bsite

-0.63

assic

-0.62

 explode

-0.62

POSITIVE LOGITS

 upholding

0.91

ably

0.89

enance

0.89

 upheld

0.81

orius

0.76

ively

0.72

 uphold

0.72

 principles

0.70

 Integrity

0.70

raints

0.69

Activations Density 0.037%

words or phrases related to legal actions and decisions

references to legal decisions or actions that affirm or maintain laws and rights

No Comments

No Known Activations