INDEX

Explanations

phrases indicating cumulative effects or consequences

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 صوتيه

-0.37

ště

-0.35

category

-0.34

 EconPapers

-0.33

Princip

-0.33

urare

-0.33

ajinan

-0.33

 Absch

-0.33

princip

-0.32

enregistrement

-0.32

POSITIVE LOGITS

 <=",

0.52

---*/

0.52

⤹

0.52

⟬

0.51

IntoConstraints

0.50

WithIOException

0.49

GraphicsUnit

0.48

fromnode

0.48

 createSlice

0.47

(~(

0.47

Activations Density 0.089%

phrases indicating cumulative effects or consequences

No Comments

No Known Activations