INDEX

Explanations

words related to physical actions, future events, and potential consequences

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

 ALWAYS

-0.68

 Everyday

-0.65

#$

-0.62

!",

-0.62

iuses

-0.62

{{

-0.61

emale

-0.60

@#

-0.60

 EVERY

-0.60

cellent

-0.60

POSITIVE LOGITS

 someday

1.28

 sooner

1.21

if

1.04

 sometime

1.03

 anytime

1.00

 elsewhere

1.00

 altogether

0.95

 soon

0.80

 nonetheless

0.78

 next

0.78

Activations Density 0.785%

words related to physical actions, future events, and potential consequences

No Comments

No Known Activations