INDEX

Explanations

references to specific names and entities, particularly related to individuals, locations, and popular culture

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

proper names and technical terms

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

asyncio

-0.40

 continuidad

-0.38

放題

-0.37

 circulación

-0.37

 corrupción

-0.35

óval

-0.35

 explotación

-0.35

 dictadura

-0.34

 adaptación

-0.34

 aporta

-0.33

POSITIVE LOGITS

AndEndTag

0.68

AutoScaleMode

0.53

 препратки

0.52

 AssemblyVersion

0.49

 snippetHide

0.49

 Audiodateien

0.49

enumii

0.48

SOUNDBITE

0.47

 beginnetje

0.47

+:+

0.47

Activations Density 0.075%

references to specific names and entities, particularly related to individuals, locations, and popular culture

proper names and technical terms

No Comments

No Known Activations