INDEX

Explanations

references to numerical margins or edges

oai_token-act-pair · gpt-3.5-turbo

references to margins in various contexts, indicating importance or comparison

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ha

-0.72

activated

-0.68

ãĥ¯ãĥ³

-0.68

HA

-0.66

ario

-0.66

olver

-0.65

heim

-0.65

race

-0.64

bre

-0.63

gru

-0.62

POSITIVE LOGITS

 margins

4.42

 margin

2.01

margin

1.52

 edges

1.42

 majorities

1.41

marg

1.31

 percentages

1.20

 corridors

1.18

 folds

1.14

 probabilities

1.11

Activations Density 0.005%

references to numerical margins or edges

references to margins in various contexts, indicating importance or comparison

No Comments

No Known Activations