INDEX

Explanations

frequent occurrences of the words related to action and participation in various contexts

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

Personendaten

-0.60

Comté

-0.54

bootstrapcdn

-0.52

 paramString

-0.51

makeConstraints

-0.50

CPtr

-0.50

LLocation

-0.49

postIndex

-0.48

IContainer

-0.47

thwaite

-0.47

POSITIVE LOGITS

 frequently

0.81

 often

0.72

frequently

0.68

 frequent

0.67

 regularly

0.66

 frequentemente

0.65

often

0.61

 spesso

0.59

Often

0.59

 frecuentemente

0.59

Activations Density 0.039%

frequent occurrences of the words related to action and participation in various contexts

No Comments

No Known Activations