INDEX

Explanations

specific terms containing the non-English word "ord."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "cord" and variations of it

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

prus

-0.67

CVE

-0.66

 midterm

-0.66

cling

-0.63

cled

-0.62

)=(

-0.62

pig

-0.60

Pwr

-0.60

 flank

-0.59

 goodness

-0.59

POSITIVE LOGITS

ination

0.99

inator

0.98

nance

0.96

inates

0.94

ord

0.94

inated

0.93

inarily

0.91

inances

0.88

elf

0.87

iev

0.86

Activations Density 0.007%

specific terms containing the non-English word "ord."

instances of the word "cord" and variations of it

No Comments

No Known Activations