INDEX

Explanations

phrases related to cost or value propositions

oai_token-act-pair · gpt-3.5-turbo

evaluative phrases indicating positive or favorable characteristics

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mares

-0.77

aline

-0.68

ansk

-0.65

Article

-0.62

alks

-0.62

etz

-0.61

images

-0.59

dule

-0.59

 rates

-0.59

individual

-0.59

POSITIVE LOGITS

 worthwhile

1.18

 boon

1.12

 viable

1.10

 good

1.07

 fruitful

1.05

 useful

1.00

 mistake

0.99

 better

0.97

 worthy

0.95

 wiser

0.95

Activations Density 0.191%

phrases related to cost or value propositions

evaluative phrases indicating positive or favorable characteristics

No Comments

No Known Activations