INDEX

Explanations

references to legislative measures or financial terms

oai_token-act-pair · gpt-3.5-turbo

parentheses in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 inward

-0.59

 guard

-0.55

 confront

-0.55

nic

-0.55

zoo

-0.55

 caution

-0.54

 cooker

-0.54

 outweigh

-0.54

tid

-0.54

 retreat

-0.53

POSITIVE LOGITS

which

1.54

including

1.52

both

1.41

such

1.33

whose

1.32

excluding

1.31

?),

1.28

perhaps

1.28

they

1.27

especially

1.27

Activations Density 0.135%

references to legislative measures or financial terms

parentheses in the text

No Comments

No Known Activations