INDEX

Explanations

statistical data and numeric representations

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

percentages and numerical values next to corresponding words or phrases

oai_token-act-pair · gpt-3.5-turbo

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

especially

-0.72

loo

-0.71

liquid

-0.69

unknown

-0.69

wered

-0.65

 Liquid

-0.64

multipl

-0.64

lay

-0.64

lu

-0.63

coll

-0.61

POSITIVE LOGITS

eenth

0.88

 cents

0.84

 percent

0.79

iffe

0.75

 anymore

0.75

Eag

0.75

 Grimm

0.72

 feet

0.69

 inches

0.68

 spots

0.67

Activations Density 0.169%

statistical data and numeric representations

percentages and numerical values next to corresponding words or phrases

No Comments

No Known Activations