INDEX

Explanations

sequences of characters that may not correspond to regular words or phrases

oai_token-act-pair · gpt-3.5-turbo

special characters or symbols typically used in technical or artistic contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 mathemat

-0.82

 Sparkle

-0.71

 relevance

-0.62

 approximate

-0.62

swick

-0.58

 recomm

-0.58

 poker

-0.57

 appointments

-0.57

 pickups

-0.57

 override

-0.56

POSITIVE LOGITS

ÏĦ

1.08

¸

0.98

ï¸ı

0.97

é¾įåĸļå£«

0.94

Ct

0.93

rh

0.93

Ã±a

0.91

ng

0.90

³

0.89

ĥ

0.88

Activations Density 0.029%

sequences of characters that may not correspond to regular words or phrases

special characters or symbols typically used in technical or artistic contexts

No Comments

No Known Activations