INDEX

Explanations

numeric hygiene rating values related to environmental and sustainable practices

oai_token-act-pair · gpt-3.5-turbo

negative sentiment or descriptions of adverse conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

vel

-0.79

 Galile

-0.72

ESC

-0.64

 Targ

-0.63

 Rodrigo

-0.63

 Vive

-0.62

Wyn

-0.62

nesday

-0.61

HOU

-0.60

 Lanka

-0.60

POSITIVE LOGITS

Specific

1.02

specific

1.02

appropriate

1.01

conscious

0.95

based

0.94

changing

0.91

selection

0.91

appropri

0.89

relation

0.88

attribute

0.87

Activations Density 0.105%

numeric hygiene rating values related to environmental and sustainable practices

negative sentiment or descriptions of adverse conditions

No Comments

No Known Activations