INDEX

Explanations

significant years or dates written in the format of two numbers separated by a dash and ending in a zero followed by other numbers

oai_token-act-pair · gpt-3.5-turbo

specific years or numerical dates in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

irlf

-0.72

lying

-0.69

 bunny

-0.68

 gren

-0.67

 peeled

-0.66

 tremend

-0.64

 hottest

-0.64

 brightest

-0.64

 holiday

-0.63

igating

-0.63

POSITIVE LOGITS

1.11

0.98

0.97

0.96

0.95

0.94

Activations Density 0.038%

significant years or dates written in the format of two numbers separated by a dash and ending in a zero followed by other numbers

specific years or numerical dates in the text

No Comments

No Known Activations