INDEX

Explanations

phrases related to retrieval and safety concerning objects or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.38.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

bootstrapcdn

-0.50

abestanden

-0.45



-0.37

GEBURTSDATUM

-0.36

ffilm

-0.36

 TextInputType

-0.36

intios

-0.35

 يتيمه

-0.35

}{@

-0.35

thyst

-0.34

POSITIVE LOGITS

 inside

3.67

 Inside

3.31

Inside

3.30

inside

3.28

 INSIDE

3.08

 internal

3.00

INSIDE

2.83

 interior

2.83

 internally

2.77

 insides

2.69

Activations Density 1.955%

phrases related to retrieval and safety concerning objects or actions

No Comments

No Known Activations