INDEX

Explanations

values and numerical data points in a variety of contexts

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

Values or errors, often numeric

np_acts-logits-general · gemini-2.0-flash

measurement error, resistance, minimal

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

AndEndTag

-0.55

ValueStyle

-0.51

########.

-0.51

 CreateTagHelper

-0.51

 surla

-0.50



-0.47

SpringRunner

-0.46

 Reverso

-0.44

ScopeManager

-0.44

IsContent

-0.43

POSITIVE LOGITS

CloseOperation

0.53

enderror

0.47

ModelSerializer

0.39

printStackTrace

0.38

unknownFields

0.37

 vanishes

0.36

 minimal

0.36

NoError

0.36

 minimized

0.35

labus

0.35

Activations Density 1.079%

values and numerical data points in a variety of contexts

Values or errors, often numeric

measurement error, resistance, minimal

No Comments

No Known Activations