INDEX

Explanations

phrases related to specific events or actions in a structured format akin to news headlines

oai_token-act-pair · gpt-3.5-turbo

references to specific events or actions tied to military or investigative themes

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

âĢ

-0.76

 âĢ

-0.73

ãĢį

-0.73

ãĢ

-0.70

AMY

-0.69

 --------

-0.68

 ðŁĳ

-0.66

âĿ

-0.65

another

-0.63

ô

-0.63

POSITIVE LOGITS

 brut

0.64

 landfill

0.60

 rubbish

0.56

 outlandish

0.55

 cramped

0.55

 impractical

0.54

 genitals

0.53

 brute

0.52

 postage

0.52

Activations Density 1.282%

phrases related to specific events or actions in a structured format akin to news headlines

references to specific events or actions tied to military or investigative themes

No Comments

No Known Activations