INDEX

Explanations

sequences of letters followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to specific entities or names

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dere

-0.66

 Idlib

-0.63

 impunity

-0.63

 Homo

-0.61

 Pharaoh

-0.60

 Gutenberg

-0.57

 laughing

-0.56

 face

-0.56

 foil

-0.56

 departure

-0.55

POSITIVE LOGITS

odies

1.22

ishop

1.22

ranch

1.20

ureau

1.16

ODY

1.15

udget

1.14

rief

1.13

ridges

1.11

irds

1.11

ooth

1.11

Activations Density 0.044%

sequences of letters followed by a number

references to specific entities or names

No Comments

No Known Activations