INDEX

Explanations

unexpected or surprising statements about various subjects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 21-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_21/width_16k/average_l0_129

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.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

 <<<<<<<<<<<<<<

-0.43

 Clearly

-0.42

 clearly

-0.42

Clearly

-0.42

 adelante

-0.41

 clairement

-0.41

farwyddwr

-0.40

 claro

-0.39

anae

-0.39

 circula

-0.39

POSITIVE LOGITS

 surprisingly

0.69

竟

0.66

 Surprisingly

0.65

居然

0.62

 unexpectedly

0.61

Surprisingly

0.59

 strangely

0.58

 oddly

0.57

竟然

0.56

なんと

0.56

Activations Density 0.454%

unexpected or surprising statements about various subjects

No Comments

No Known Activations