INDEX

Explanations

mathematical symbols and codes, potentially related to a specific technical or scientific context

oai_token-act-pair · gpt-3.5-turbo

symbols or characters with high activation values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

uld

-0.91

wcs

-0.91

stro

-0.87

enos

-0.84

ulates

-0.80

iqueness

-0.79

gow

-0.78

ulate

-0.78

igree

-0.78

eni

-0.77

POSITIVE LOGITS

âĪ

0.95

âĶĢâĶĢ

0.89

âĢ¢âĢ¢âĢ¢âĢ¢

0.86

×Ļ×

0.85

 âĪ

0.84

ãĥĺ

0.83

©

0.80

ROR

0.80

ãĥķãĤ©

0.79

§

0.77

Activations Density 0.009%

mathematical symbols and codes, potentially related to a specific technical or scientific context

symbols or characters with high activation values

No Comments

No Known Activations