INDEX

Explanations

numerical data presented with a comma

oai_token-act-pair · gpt-3.5-turbo

patterns of statistical data and numerical comparisons in a text

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

²¾

-0.82

¬¼

-0.71

obi

-0.71

tnc

-0.66

igm

-0.64

dden

-0.63

oner

-0.61

Ĥİ

-0.61

oslav

-0.61

enson

-0.59

POSITIVE LOGITS

 compared

1.31

 whereas

1.22

 according

1.21

according

1.00

 while

0.98

 versus

0.97

 trailed

0.95

 indicating

0.94

 implying

0.92

 representing

0.92

Activations Density 0.191%

numerical data presented with a comma

patterns of statistical data and numerical comparisons in a text

No Comments

No Known Activations