INDEX

Explanations

occurrences of the word "each" and other variants related to quantifying or specifying items in a list

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

 hayas

-0.38

UserContext

-0.34

 разве

-0.34

kubectl

-0.34

Hint

-0.33

SetName

-0.33

 tidak

-0.33

isez

-0.32

 ను

-0.32

 espere

-0.32

POSITIVE LOGITS

AndEndTag

0.75

 tiap

0.73

 each

0.73

 Ogni

0.68

 individual

0.68

Each

0.67

每個

0.66

 Each

0.65

 Every

0.65

usercontent

0.65

Activations Density 0.061%

occurrences of the word "each" and other variants related to quantifying or specifying items in a list

No Comments

No Known Activations