INDEX

Explanations

negative descriptors or attributes

oai_token-act-pair · gpt-3.5-turbo

instances of the word "bad" and its associations with negative outcomes or 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.96

theless

-0.81

ittees

-0.80

 Pavilion

-0.75

earchers

-0.74

olate

-0.73

eston

-0.70

ymph

-0.70

uine

-0.69

eters

-0.69

POSITIVE LOGITS

dies

1.14

dest

1.07

 luck

1.06

ger

1.06

die

1.05

gered

1.04

 karma

1.01

asses

0.94

 publicity

0.92

der

0.90

Activations Density 0.040%

negative descriptors or attributes

instances of the word "bad" and its associations with negative outcomes or conditions

No Comments

No Known Activations