INDEX

Explanations

numerical values followed by units of time or distance

oai_token-act-pair · gpt-3.5-turbo

numeric values, particularly those related to time, quantity, and percentages

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

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

 Caption

-0.70

 guid

-0.66

 Kimmel

-0.66

actionDate

-0.63

 attracting

-0.60

 Greenwald

-0.58

ãĤ¦

-0.58

 targeting

-0.57

 Tradable

-0.57

igation

-0.57

POSITIVE LOGITS

teenth

0.96

aciously

0.87

 years

0.87

ecause

0.86

 bucks

0.82

 cents

0.81

teen

0.81

+.

0.78

+)

0.78

years

0.77

Activations Density 0.390%

numerical values followed by units of time or distance

numeric values, particularly those related to time, quantity, and percentages

No Comments

No Known Activations