INDEX

Explanations

instances of the word "the."

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

numbers and years

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

 ब्रेकडाउन

-0.82

 myſelf

-0.71



-0.66

 <>",

-0.66

 vägen

-0.64

WithIOException

-0.64

IntoConstraints

-0.63

 AssemblyTitle

-0.63

 propOrder

-0.60

 Chwiliwch

-0.58

POSITIVE LOGITS

va

0.45

ask

0.43

ga

0.43

vo

0.42

di

0.42

pe

0.41

vas

0.40

Mun

0.40

So

0.40

mun

0.39

Activations Density 0.115%

instances of the word "the."

numbers and years

No Comments

No Known Activations