INDEX

Explanations

punctuation and conjunctions in various contexts

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

user turn indicators

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

 betweenstory

-0.61

PostInfinity

-0.60

PerformLayout

-0.49

RegressionTest

-0.48

Tari

-0.47



-0.46

 BorderSide

-0.46

expandindo

-0.45

 OMITBAD

-0.45

OGND

-0.44

POSITIVE LOGITS

NUMX

0.41

complexContent

0.37

PREFERRED

0.37

Chham

0.36

BeginContext

0.35

WebServlet

0.33

MemoryWarning

0.33

 consultato

0.33

BuildContext

0.32

makeConstraints

0.32

Activations Density 0.001%

punctuation and conjunctions in various contexts

user turn indicators

No Comments

No Known Activations