INDEX

Explanations

terms related to semantics and structures implying organization

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

loss of semantics or relations

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

<eos>

-0.37

ท

-0.29

 prévoir

-0.29

 policial

-0.28

 éviter

-0.28

 Irán

-0.27

 masculino

-0.27

 logra

-0.27

 Watch

-0.27

POSITIVE LOGITS

semantics

4.75

 Semantics

1.73

 semantics

1.66

Semantics

1.65

 seman

1.02

semantic

1.00

seman

1.00

ValueStyle

0.99

awtextra

0.98

 semantic

0.96

Activations Density 0.000%

terms related to semantics and structures implying organization

loss of semantics or relations

No Comments

No Known Activations