INDEX

Explanations

references to specific episodes and their details

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 للمعارف

-0.74

 CreateTagHelper

-0.59

 BoxDecoration

-0.50

DockStyle

-0.48

 démocr

-0.47

 constitución

-0.46

SharedCtor

-0.45

rungsseite

-0.43

 bezit

-0.42

protetor

-0.41

POSITIVE LOGITS

 episode

0.77

 episodes

0.73

episode

0.64

Episode

0.61

 episodi

0.60

Epis

0.58

ep

0.57

eps

0.56

 Episode

0.55

pisode

0.53

Activations Density 0.043%

references to specific episodes and their details

No Comments

No Known Activations