INDEX

Explanations

words related to intention or purpose

oai_token-act-pair · gpt-3.5-turbo

phrases indicating purpose or intent

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

uesday

-0.70

natureconservancy

-0.60

reports

-0.54

 Transcript

-0.53

 Flavoring

-0.52

 Mahjong

-0.52

urus

-0.51

ua

-0.51

Zen

-0.50

Reports

-0.50

POSITIVE LOGITS

to

0.95

 solely

0.90

 primarily

0.87

 purely

0.84

 principally

0.82

for

0.80

 specifically

0.74

to

0.73

 chiefly

0.71

 exclusively

0.71

Activations Density 0.067%

words related to intention or purpose

phrases indicating purpose or intent

No Comments

No Known Activations