INDEX

Explanations

occurrences of the word "and" in various contexts

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

connecting clauses with and

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

amate

-0.39

ktm

-0.38

amd

-0.36

UNRELATED

-0.36

 pain

-0.36

 care

-0.36

 mixed

-0.35

bag

-0.35

Bow

-0.35

Vit

-0.35

POSITIVE LOGITS

 inoltre

0.66

 estekak

0.65

 demikian

0.57

 ayrıca

0.56

 dessutom

0.52

 pertanto

0.51

 ligiloj

0.49

 Walkover

0.49

 moreover

0.48

anjutnya

0.48

Activations Density 0.059%

occurrences of the word "and" in various contexts

connecting clauses with and

No Comments

No Known Activations