INDEX

Explanations

strings of text related to mentions of different companies, organizations, or individuals

oai_token-act-pair · gpt-3.5-turbo

references to public figures or events in a formal context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

-1.13

defense

-0.92

bors

-0.89

 canceled

-0.87

caliber

-0.86

rollment

-0.84

avored

-0.84

 honoring

-0.81

 favors

-0.76

 flavorful

-0.75

POSITIVE LOGITS

 Whilst

1.56

 realise

1.52

 Â£

1.51

 recognised

1.48

UK

1.47

 whilst

1.45

 organis

1.45

NZ

1.44

 organised

1.44

 organise

1.43

Activations Density 1.659%

strings of text related to mentions of different companies, organizations, or individuals

references to public figures or events in a formal context

No Comments

No Known Activations