INDEX

Explanations

phrases related to completion and remaining amounts in various contexts

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

remaining things

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

 television

-0.36

 adanya

-0.35

TV

-0.33

Therefore

-0.33

)))),

-0.33

gan

-0.33

 abbia

-0.32

 directly

-0.31

typeparam

-0.31

很赞哦

-0.31

POSITIVE LOGITS

 Remaining

0.66

OGND

0.65

 restantes

0.62

jsonwebtoken

0.61

DockStyle

0.60

 المعيارى

0.59

 remaining

0.59

devamını

0.58

出版年

0.58

Remaining

0.57

Activations Density 0.037%

phrases related to completion and remaining amounts in various contexts

remaining things

No Comments

No Known Activations