INDEX

Explanations

proper titles and abbreviations related to doctors

oai_token-act-pair · gpt-3.5-turbo

references to characters with the title "Dr." followed by a name

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 administr

-0.68

words

-0.66

 coupling

-0.66

 counsel

-0.65

 compe

-0.64

 colle

-0.64

alities

-0.63

amy

-0.63

 volley

-0.62

 duly

-0.62

POSITIVE LOGITS

 Toad

0.88

Marvel

0.86

 Crusher

0.85

Planet

0.84

 Boom

0.82

 Miracle

0.82

 Freeze

0.82

 Feel

0.82

 Universe

0.80

 Pump

0.79

Activations Density 0.045%

proper titles and abbreviations related to doctors

references to characters with the title "Dr." followed by a name

No Comments

No Known Activations