INDEX

Explanations

terms related to legal cases and their components

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

contralateral hemisphere aspect expected negative

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

 purpoſe

-0.46

 leſs

-0.45

 uſe

-0.43

 uſed

-0.42

 tranſ

-0.42

 pleaſure

-0.42

 ſeveral

-0.41

 diſt

-0.40

 ſur

-0.39

 raiſ

-0.38

POSITIVE LOGITS

 AssemblyTitle

0.45

 segnale

0.44

 SEDS

0.41

ISupport

0.39

 Irrelevant

0.38

GMENT

0.38

 informée

0.38

 gyhoeddwyd

0.38

 CreateTagHelper

0.38

 derde

0.37

Activations Density 0.003%

terms related to legal cases and their components

contralateral hemisphere aspect expected negative

No Comments

No Known Activations