INDEX

Explanations

information about news articles and journalistic content

oai_token-act-pair · gpt-3.5-turbo

references to legal or judicial proceedings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

colored

-1.06

defense

-0.91

avored

-0.91

bors

-0.89

caliber

-0.85

Defense

-0.84

 favorably

-0.83

 honor

-0.82

 favors

-0.82

 gotten

-0.82

POSITIVE LOGITS

 Â£

1.59

 recognise

1.53

 recognised

1.47

 realise

1.39

 programme

1.38

 maths

1.36

 centre

1.35

 organising

1.33

 organised

1.33

 analyse

1.33

Activations Density 4.495%

information about news articles and journalistic content

references to legal or judicial proceedings

No Comments

No Known Activations