INDEX

Explanations

punctuation marks and their context within programming or code-related discussions

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

answer following "A:"

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

WriteBarrier

-0.56

Diweddarwch

-0.42

 跳转至

-0.40

ManagedBean

-0.38

 protoimpl

-0.38

 jadx

-0.38

сылкі

-0.36

Verdana

-0.36

nodoc

-0.36

jshint

-0.35

POSITIVE LOGITS

 للمعارف

0.48

 autorytatywna

0.44

 veuillez

0.42

 Picchu

0.41

 defaultstate

0.41

osamente

0.41

ablemente

0.40

setVerticalGroup

0.39

 GenerationType

0.38

RTEX

0.38

Activations Density 0.113%

punctuation marks and their context within programming or code-related discussions

answer following "A:"

No Comments

No Known Activations