INDEX

Explanations

names of places and organizations

oai_token-act-pair · gpt-3.5-turbo

specific abbreviations or acronyms commonly used in specific contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ngth

-1.01

ãĥĩãĤ£

-0.71

£ı

-0.68

retty

-0.66

ãĤ¼ãĤ¦ãĤ¹

-0.66

ãĤ©

-0.65

ãĥĻ

-0.64

ãĤ¨ãĥ«

-0.64

ufact

-0.64

ãĥ¢

-0.64

POSITIVE LOGITS

bush

0.72

anski

0.69

spir

0.68

Tip

0.68

hu

0.66

ĪĴ

0.65

Tel

0.62

arat

0.61

aq

0.61

ihu

0.61

Activations Density 0.090%

names of places and organizations

specific abbreviations or acronyms commonly used in specific contexts

No Comments

No Known Activations