INDEX

Explanations

sequences that resemble timestamps or coded data patterns

oai_token-act-pair · gpt-3.5-turbo

punctuation marks, particularly colons

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

 mathemat

-0.76

 ratified

-0.70

 fatally

-0.68

steen

-0.67

 bounded

-0.64

GBT

-0.63

 paralyzed

-0.63

 incent

-0.63

 Chambers

-0.62

vier

-0.62

POSITIVE LOGITS

</

0.79

)</

0.77

addons

0.73

 Holo

0.70

\">

0.70

/>

0.70

Display

0.69

)\

0.69

antage

0.68

ãĢį

0.67

Activations Density 0.023%

sequences that resemble timestamps or coded data patterns

punctuation marks, particularly colons

No Comments

No Known Activations