INDEX

Explanations

combinations of numbers and letters that don't form meaningful words or sentences

oai_token-act-pair · gpt-3.5-turbo

numeric data and identifiers

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 encour

-0.61

 cynical

-0.60

 editorial

-0.60

 tentative

-0.60

 Ober

-0.58

 indications

-0.58

 mathemat

-0.58

 roadside

-0.57

 grappling

-0.57

 Cecil

-0.57

POSITIVE LOGITS

tx

0.92

}}

0.83

"}

0.81

mask

0.80

sac

0.79

Î¼

0.78

icon

0.77

xx

0.77

})

0.77

node

0.77

Activations Density 0.183%

combinations of numbers and letters that don't form meaningful words or sentences

numeric data and identifiers

No Comments

No Known Activations