INDEX

Explanations

patterns related to timestamps or structured data formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 13-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_13/width_65k/average_l0_75

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.13.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MemoryWarning

-0.63

 protoimpl

-0.49

izie

-0.47

<eos>

-0.47

ioutil

-0.45

AutoScaleMode

-0.44

OrNil

-0.44

 createState

-0.43

 brief

-0.43

zzino

-0.42

POSITIVE LOGITS

 houſe

0.71

 ſtate

0.66

 myſelf

0.63

 Houſe

0.63

 purpoſe

0.63

 obfer

0.61

 Majefty

0.59

 clín

0.59

licates

0.58

 faſt

0.58

Activations Density 0.004%

patterns related to timestamps or structured data formats

No Comments

No Known Activations