INDEX

Explanations

numbers reflecting timelines or durations

oai_token-act-pair · gpt-3.5-turbo

numerical values and time durations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

NAS

-0.91

lance

-0.80

Duo

-0.78

Joy

-0.75

vale

-0.71

MSN

-0.70

-0.67

alogy

-0.67

 Prelude

-0.66

acus

-0.66

POSITIVE LOGITS

ths

0.97

 thousand

0.95

 percent

0.93

 dozen

0.92

 times

0.92

 thirds

0.86

 hundred

0.85

 million

0.84

 generations

0.83

 years

0.81

Activations Density 0.133%

numbers reflecting timelines or durations

numerical values and time durations

No Comments

No Known Activations