INDEX

Explanations

specific actions or choices related to personal preferences

oai_token-act-pair · gpt-3.5-turbo

elements related to choices and actions 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 @ 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

cknowled

-0.57

 millenn

-0.57

 Freak

-0.57

ueless

-0.56

ady

-0.55

 cumbersome

-0.54

 willful

-0.54

ĸļ

-0.54

 Beverly

-0.54

¥µ

-0.53

POSITIVE LOGITS

 depends

0.98

 versus

0.87

 determines

0.86

 next

0.81

 beforehand

0.81

vs

0.77

 based

0.75

 closest

0.75

?,

0.73

,...

0.73

Activations Density 0.362%

specific actions or choices related to personal preferences

elements related to choices and actions in various contexts

No Comments

No Known Activations