INDEX

Explanations

frequency of numerical values and operational commands in a technical context

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

code snippets or variable assignments

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

Negative Logits

<unused41>

-1.07

<unused28>

-1.06

<unused52>

-1.06

<unused23>

-1.06

<unused74>

-1.06

<unused8>

-1.06

<unused79>

-1.06

<unused14>

-1.05

[@BOS@]

-1.05

<pad>

-1.05

POSITIVE LOGITS

 hatta

0.41

But

0.38

0.34

Và

0.34

and

0.33

 valamint

0.33

Why

0.33

etc

0.32

 What

0.31

And

0.30

Activations Density 0.107%

frequency of numerical values and operational commands in a technical context

code snippets or variable assignments

No Known Activations