INDEX

Explanations

the word element "ib" followed by a single-digit numerical value (ranging from 8 to 10)

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 bothering

-0.61

CRIP

-0.60

PAC

-0.59

Ain

-0.59

 merit

-0.58

 admitting

-0.57

CHO

-0.57

 boarding

-0.57

 rejecting

-0.57

 Furious

-0.57

POSITIVE LOGITS

raltar

1.29

bons

1.28

bler

1.24

rahim

1.17

bles

1.14

ilant

1.12

bon

1.08

iotics

1.06

odies

1.06

acter

1.06

Activations Density 0.042%

the word element "ib" followed by a single-digit numerical value (ranging from 8 to 10)

No Comments

No Known Activations