INDEX

Explanations

phrases related to software patches or updates

oai_token-act-pair · gpt-3.5-turbo

references to software patches

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

yss

-0.74

emouth

-0.68

amic

-0.67

cius

-0.64

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.64

agy

-0.62

minist

-0.62

¢

-0.62

relative

-0.62

uron

-0.61

POSITIVE LOGITS

work

0.94

 patched

0.92

 patch

0.89

notes

0.88

ions

0.83

ioned

0.79

 patches

0.75

 Patch

0.72

ing

0.71

boxes

0.69

Activations Density 0.012%

phrases related to software patches or updates

references to software patches

No Comments

No Known Activations