INDEX

Explanations

references to ceremonies or rituals

oai_token-act-pair · gpt-3.5-turbo

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 delay

-0.69

 raft

-0.63

 Izan

-0.62

enegger

-0.61

 Fiorina

-0.61

 Bastard

-0.61

 Chains

-0.60

 numbered

-0.60

 Hurricanes

-0.59

 skies

-0.58

POSITIVE LOGITS

quez

1.25

zzle

1.20

quet

1.19

zona

1.13

ving

1.12

vals

1.09

ety

1.08

zz

1.03

archy

1.00

elsen

1.00

Activations Density 0.033%

references to ceremonies or rituals

No Comments

No Known Activations