INDEX

Explanations

words or phrases highlighted with special characters Ċ and Ċ

oai_token-act-pair · gpt-3.5-turbo

instances of various numerical values or metrics

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

etheless

-1.00

bda

-0.78

conservancy

-0.77

 composing

-0.74

 conscious

-0.72

 bask

-0.72

 overloaded

-0.72

 endeav

-0.69

 filtering

-0.68

 periodic

-0.68

POSITIVE LOGITS

SPONSORED

1.06

Newsletter

1.01

Said

0.95

Official

0.93

Wilson

0.92

Lev

0.85

ONSORED

0.83

["

0.83

Myth

0.83

Reward

0.82

Activations Density 0.217%

words or phrases highlighted with special characters Ċ and Ċ

instances of various numerical values or metrics

No Comments

No Known Activations