INDEX

Explanations

phrases indicating participation or involvement

oai_token-act-pair · gpt-3.5-turbo

references to being "on board" or included in a group or initiative

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

bered

-0.85

byter

-0.66

PLE

-0.65

olver

-0.65

ously

-0.64

ascus

-0.63

usalem

-0.63

çīĪ

-0.63

sg

-0.61

Dum

-0.61

POSITIVE LOGITS

walk

1.27

wal

0.95

deck

0.94

meal

0.87

mast

0.81

boats

0.81

boat

0.79

mates

0.78

board

0.78

chairs

0.74

Activations Density 0.013%

phrases indicating participation or involvement

references to being "on board" or included in a group or initiative

No Comments

No Known Activations