INDEX

Explanations

mentions of law enforcement actions or legal terms

oai_token-act-pair · gpt-3.5-turbo

references to police and legal situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Secondly

-0.51

âĹ¼

-0.50

 weaker

-0.46

 successors

-0.45

sic

-0.44

pse

-0.44

 aforementioned

-0.43

 distant

-0.43

 weak

-0.42

 disob

-0.42

POSITIVE LOGITS

ashtra

0.58

 âĢº

0.56

 SHARES

0.54

undrum

0.52

allery

0.50

 Calculator

0.50

ategor

0.50

 Volunte

0.50

ocument

0.50

hya

0.49

Activations Density 1.526%

mentions of law enforcement actions or legal terms

references to police and legal situations

No Comments

No Known Activations