INDEX

Explanations

phrases related to particular actions or approaches

oai_token-act-pair · gpt-3.5-turbo

phrases indicating direction or approach towards something

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DragonMagazine

-0.96

 teasp

-0.85

miah

-0.84

aldi

-0.81

RIP

-0.78

minus

-0.78

ciation

-0.78

ographer

-0.73

LAN

-0.72

Scroll

-0.69

POSITIVE LOGITS

 dealing

1.57

 addressing

1.52

 tackling

1.49

 combating

1.45

 discussing

1.43

 assessing

1.41

 interpreting

1.38

 promoting

1.37

 evaluating

1.37

 detecting

1.37

Activations Density 0.225%

phrases related to particular actions or approaches

phrases indicating direction or approach towards something

No Comments

No Known Activations