INDEX

Explanations

mentions of "RS" or acronyms containing "RS" with a high activation value

oai_token-act-pair · gpt-3.5-turbo

references to rankings or scores associated with events or entities

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

 villain

-0.63

 Falling

-0.61

nan

-0.59

 dove

-0.58

 plum

-0.58

 melting

-0.57

 Uzbek

-0.57

 Babylon

-0.57

 Elaine

-0.57

 pill

-0.57

POSITIVE LOGITS

RS

4.42

RS

1.93

RL

1.61

RC

1.60

rs

1.50

FS

1.39

RR

1.38

DS

1.37

RD

1.34

WS

1.34

Activations Density 0.013%

mentions of "RS" or acronyms containing "RS" with a high activation value

references to rankings or scores associated with events or entities

No Comments

No Known Activations