INDEX

Explanations

numerical equality and comparisons

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

asel

-0.85

hari

-0.72

beh

-0.71

avorite

-0.65

é¾įå

-0.64

oji

-0.62

Beh

-0.61

aptic

-0.61

stal

-0.61

////////////////

-0.60

POSITIVE LOGITS

 proportions

0.80

inity

0.77

izes

0.75

amount

0.73

ized

0.73

 TOTAL

0.72

ivalent

0.72

MPG

0.69

izing

0.69

Amount

0.68

Activations Density 0.015%

numerical equality and comparisons

No Comments

No Known Activations