INDEX

Explanations

instances of the word "ek"

oai_token-act-pair · gpt-3.5-turbo

references to a specific name or term repeated multiple times

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 AMERICA

-0.67

 Bland

-0.64

âĢ¢âĢ¢âĢ¢âĢ¢

-0.62

 cort

-0.61

appropriately

-0.60

ãĥĩãĤ£

-0.60

 Continental

-0.59

ENTS

-0.57

 Hussain

-0.57

 diapers

-0.57

POSITIVE LOGITS

nown

1.55

enstein

1.06

hov

1.00

ernel

0.97

kers

0.97

yll

0.94

anism

0.93

anan

0.93

enzie

0.93

kies

0.93

Activations Density 0.047%

instances of the word "ek"

references to a specific name or term repeated multiple times

No Comments

No Known Activations