INDEX

Explanations

words related to commands, recommendations, or procedures

oai_token-act-pair · gpt-3.5-turbo

words related to commendation and commendable actions

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

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

 dece

-0.65

 Vietnamese

-0.63

 Feldman

-0.61

 Balk

-0.60

teen

-0.59

chers

-0.59

Bav

-0.59

 Wolfgang

-0.58

liest

-0.57

 Finnish

-0.57

POSITIVE LOGITS

encing

1.41

itte

1.37

ittal

1.29

otion

1.26

issions

1.25

ences

1.24

ittee

1.24

ittees

1.20

utation

1.20

onsense

1.17

Activations Density 0.022%

words related to commands, recommendations, or procedures

words related to commendation and commendable actions

No Comments

No Known Activations