INDEX

Explanations

names of researchers, editors, and authors in academic contexts

oai_token-act-pair · gpt-3.5-turbo

plural nouns or nouns with similar endings

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

 exception

-0.62

 exceptions

-0.59

 Ocean

-0.58

 fraction

-0.58

 recharge

-0.56

 exemptions

-0.56

 comparable

-0.55

SIGN

-0.55

ASED

-0.55

 arts

-0.55

POSITIVE LOGITS

ki

1.83

kaya

1.75

ky

1.72

mith

1.69

hire

1.57

nyder

1.47

hip

1.45

iewicz

1.36

haw

1.36

cu

1.35

Activations Density 0.181%

names of researchers, editors, and authors in academic contexts

plural nouns or nouns with similar endings

No Comments

No Known Activations