INDEX

Explanations

sequences of characters that seem to be part of a different language or encoding

oai_token-act-pair · gpt-3.5-turbo

high-frequency characters or symbols from a non-Latin script

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

aneers

-0.78

ierrez

-0.70

ographically

-0.68

anwhile

-0.66

ortium

-0.65

theless

-0.65

ividual

-0.62

DragonMagazine

-0.62

 Perkins

-0.61

essa

-0.61

POSITIVE LOGITS

²

1.93

´

1.86

¹

1.77

Ń

1.77

º

1.77

¼

1.76

¶

1.76

·

1.74

¦

1.72

£

1.71

Activations Density 0.009%

sequences of characters that seem to be part of a different language or encoding

high-frequency characters or symbols from a non-Latin script

No Comments

No Known Activations