INDEX

Explanations

the word "pardon" and related terms

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of pardon or forgiveness

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eled

-0.68

 blob

-0.68

pled

-0.64

elled

-0.62

td

-0.61

jan

-0.61

oted

-0.61

 uninsured

-0.60

 warehouses

-0.60

arist

-0.60

POSITIVE LOGITS

ardon

1.65

 pardon

0.89

 "$:/

0.87

0.85

abouts

0.81

 pard

0.80

alty

0.78

utical

0.77

ä»£

0.75

MENTS

0.75

Activations Density 0.006%

the word "pardon" and related terms

terms related to the concept of pardon or forgiveness

No Comments

No Known Activations