INDEX

Explanations

positively-valenced descriptive terms or expressions related to moral attributes

oai_token-act-pair · gpt-3.5-turbo

mentions of positive attributes or concepts, particularly those associated with "Good" and "House."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

USS

-0.80

jab

-0.73

ULAR

-0.66

ulas

-0.65

otom

-0.65

REF

-0.62

 veins

-0.62

ceivable

-0.61

cious

-0.61

execute

-0.61

POSITIVE LOGITS

 outweigh

1.03

 outwe

1.00

 Samar

0.94

 luck

0.85

bye

0.78

nered

0.76

+++

0.74

 rewarded

0.73

 reputation

0.70

 prosper

0.68

Activations Density 0.196%

positively-valenced descriptive terms or expressions related to moral attributes

mentions of positive attributes or concepts, particularly those associated with "Good" and "House."

No Comments

No Known Activations