INDEX

Explanations

time-related phrases, specifically mentioning the number of years

oai_token-act-pair · gpt-3.5-turbo

numerical data and references to time periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ourcing

-0.67

%);

-0.65

ernels

-0.64

utor

-0.64

enegger

-0.63

uez

-0.61

othy

-0.59

FORMATION

-0.58

,,,,,,,,

-0.58

oidal

-0.58

POSITIVE LOGITS

 history

1.00

history

0.92

 decades

0.90

 years

0.88

 generations

0.87

 memory

0.83

decade

0.82

years

0.81

ann

0.76

 tries

0.74

Activations Density 0.042%

time-related phrases, specifically mentioning the number of years

numerical data and references to time periods

No Comments

No Known Activations