INDEX

Explanations

comparisons using "than" phrases

oai_token-act-pair · gpt-3.5-turbo

comparative phrases that emphasize a degree of intensity or extremity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 largeDownload

-0.65

ilic

-0.62

igible

-0.62

uty

-0.61

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.59

yna

-0.58

iga

-0.58

sonian

-0.58

aird

-0.58

EMENT

-0.58

POSITIVE LOGITS

 anything

0.76

 anybody

0.71

 usual

0.70

 others

0.68

 anyone

0.67

 anywhere

0.62

 ever

0.62

 bothering

0.61

 else

0.60

any

0.60

Activations Density 0.082%

comparisons using "than" phrases

comparative phrases that emphasize a degree of intensity or extremity

No Comments

No Known Activations