INDEX

Explanations

character names within quotes

oai_token-act-pair · gpt-3.5-turbo

quotation marks

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

 barr

-0.75

 matter

-0.75

fry

-0.70

 frying

-0.69

 express

-0.69

 prey

-0.68

dye

-0.68

 Nieto

-0.68

 favour

-0.67

 pine

-0.67

POSITIVE LOGITS

BuyableInstoreAndOnline

1.14

Untitled

1.12

Operation

1.10

Big

1.10

Where

1.08

WHERE

1.07

every

1.06

Star

1.06

Never

1.06

Dear

1.05

Activations Density 0.106%

character names within quotes

quotation marks

No Comments

No Known Activations