INDEX

Explanations

phrases related to interactions and actions involving asking, proposing, feeling, mocking, contacting, and giving

oai_token-act-pair · gpt-3.5-turbo

actions and accusations related to interpersonal conflict or controversy

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

joice

-0.80

road

-0.80

%%

-0.78

hill

-0.77

cano

-0.77

Brexit

-0.77

agate

-0.75

iHUD

-0.74

rift

-0.74

aspx

-0.71

POSITIVE LOGITS

 himself

1.47

his

1.24

 subordinates

1.01

 girlfriend

1.01

 teammates

0.94

 teammate

0.92

 Himself

0.91

 fellow

0.90

 colleagues

0.88

 others

0.88

Activations Density 0.783%

phrases related to interactions and actions involving asking, proposing, feeling, mocking, contacting, and giving

actions and accusations related to interpersonal conflict or controversy

No Comments

No Known Activations