INDEX

Explanations

numerical margins or errors mentioned in a document

oai_token-act-pair · gpt-3.5-turbo

phrases related to measurement and error margins

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

mine

-0.73

JECT

-0.70

Gil

-0.70

ibur

-0.66

Uz

-0.66

Rod

-0.65

wagen

-0.65

den

-0.62

ergy

-0.62

yer

-0.61

POSITIVE LOGITS

 margin

1.22

 margins

1.10

asonic

0.82

margin

0.79

ageddon

0.78

olicy

0.78

 unbeliev

0.74

 separating

0.72

 heights

0.70

acies

0.70

Activations Density 0.014%

numerical margins or errors mentioned in a document

phrases related to measurement and error margins

No Comments

No Known Activations