INDEX

Explanations

phrases related to interactions and relationships in various contexts

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

connecting multiple concepts

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Wern

-0.42

gien

-0.40

inth

-0.40

度

-0.40

IFI

-0.40

Apo

-0.40

fekt

-0.40

Apo

-0.40

Was

-0.39

icius

-0.39

POSITIVE LOGITS

OGND

0.78

featureID

0.76

 Савезне

0.71

WithIOException

0.59

posedge

0.58

webElementXpaths

0.57

 PeEnEo

0.56

AutoScaleMode

0.55

 ")");

0.54

 ब्रेकडाउन

0.54

Activations Density 0.084%

phrases related to interactions and relationships in various contexts

connecting multiple concepts

No Comments

No Known Activations