INDEX

Explanations

proper nouns or names of people in images or text

oai_token-act-pair · gpt-3.5-turbo

references to notable individuals or figures, particularly in the context of images or photographs

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

iencies

-0.69

cipled

-0.64

bin

-0.63

ente

-0.63

imester

-0.62

Attempts

-0.58

Marginal

-0.58

 actionGroup

-0.57

psc

-0.57

 pretext

-0.57

POSITIVE LOGITS

 smiling

0.86

 kneeling

0.83

 interacting

0.77

 grinning

0.77

 disappearing

0.72

hoe

0.71

 behaving

0.70

Done

0.70

 handwriting

0.70

 lying

0.69

Activations Density 0.699%

proper nouns or names of people in images or text

references to notable individuals or figures, particularly in the context of images or photographs

No Comments

No Known Activations