INDEX

Explanations

phrases containing the word "Pu" with varying activation levels

oai_token-act-pair · gpt-3.5-turbo

references to the name "Pu" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Cosponsors

-1.06

 GOODMAN

-0.83

hips

-0.80

rawdownloadcloneembedreportprint

-0.78

Interstitial

-0.76

*/(

-0.71

 Corinth

-0.69

 Borders

-0.67

 Beir

-0.67

 abdom

-0.67

POSITIVE LOGITS

Pu

0.99

ente

0.97

ja

0.96

issance

0.95

etooth

0.92

asa

0.92

pport

0.91

pee

0.91

cci

0.90

isine

0.89

Activations Density 0.008%

phrases containing the word "Pu" with varying activation levels

references to the name "Pu" and its variations in different contexts

No Comments

No Known Activations