INDEX

Explanations

proper nouns related to jail or correctional facilities

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals and locations associated with incarceration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

arted

-0.80

 Cumber

-0.80

 dred

-0.72

 trop

-0.72

trop

-0.70

 Chero

-0.69

olition

-0.69

 partisans

-0.67

Brook

-0.67

ply

-0.64

POSITIVE LOGITS

Sz

1.70

Ign

1.69

 Jail

1.56

Ign

1.19

conn

1.02

 Paste

0.99

Lab

0.98

****

0.96

 Alvarez

0.96

 Liam

0.94

Activations Density 0.046%

proper nouns related to jail or correctional facilities

references to specific individuals and locations associated with incarceration

No Comments

No Known Activations