INDEX

Explanations

names of individuals involved in various activities or events

oai_token-act-pair · gpt-3.5-turbo

references to individuals involved in various incidents or actions

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

 Trends

-0.59

isSpecialOrderable

-0.59

 renaissance

-0.58

 Favor

-0.57

 Rankings

-0.56

 drought

-0.55

Born

-0.54

igmatic

-0.54

 Recent

-0.54

Coming

-0.53

POSITIVE LOGITS

 then

1.61

 replied

1.35

then

1.31

 complied

1.31

 subsequently

1.30

 THEN

1.28

 countered

1.26

 proceeded

1.25

 responded

1.22

 eventually

1.18

Activations Density 0.405%

names of individuals involved in various activities or events

references to individuals involved in various incidents or actions

No Comments

No Known Activations