INDEX

Explanations

phrases related to specific actions being carried out

oai_token-act-pair · gpt-3.5-turbo

instances of the word "conducting" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

$$$$

-0.68

IER

-0.62

ury

-0.62

ikers

-0.60

XX

-0.59

 nods

-0.58

em

-0.58

HL

-0.57

axe

-0.57

 fades

-0.57

POSITIVE LOGITS

 conducting

3.42

conduct

2.00

 conducts

1.97

 conduct

1.82

 administering

1.72

 Conduct

1.63

 interviewing

1.63

 investigating

1.57

 performing

1.54

 conducted

1.52

Activations Density 0.009%

phrases related to specific actions being carried out

instances of the word "conducting" in various contexts

No Comments

No Known Activations