INDEX

Explanations

addresses with street names and cities

oai_token-act-pair · gpt-3.5-turbo

specific addresses and locations

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

 largeDownload

-0.86

=-=-=-=-=-=-=-=-

-0.74

 databases

-0.71

ãģĦ

-0.70

icter

-0.69

BACK

-0.69

ãģ£

-0.68

ãĤ§

-0.67

 programmes

-0.66

 sums

-0.66

POSITIVE LOGITS

 Blvd

1.01

 Amsterdam

0.93

NW

0.85

 Suite

0.85

 Avalon

0.82

 Macy

0.79

 Marriott

0.78

 Boulevard

0.77

 Pleasant

0.76

Ave

0.76

Activations Density 0.088%

addresses with street names and cities

specific addresses and locations

No Comments

No Known Activations