INDEX

Explanations

specific mentions of the abbreviation "CP" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

references to polling or survey results

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

Goo

-0.77

sbm

-0.71

ebted

-0.68

âĸ¬

-0.67

ãĥĦ

-0.65

 Admir

-0.64

geons

-0.64

wagen

-0.63

tainment

-0.63

iyah

-0.62

POSITIVE LOGITS

ython

1.01

ople

0.93

erity

0.92

etition

0.86

illon

0.86

ollo

0.86

ocalypse

0.85

resent

0.84

ropri

0.84

AN

0.83

Activations Density 0.011%

specific mentions of the abbreviation "CP" with a high level of activation

references to polling or survey results

No Comments

No Known Activations