INDEX

Explanations

patterns related to numerical data and codes

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

utilisons

-0.60

TagMode

-0.55

 ſta

-0.52

 pleaſure

-0.51

 houſe

-0.51

BeginContext

-0.50

 beſ

-0.50

 Reſ

-0.50

 unſ

-0.49

 ſche

-0.49

POSITIVE LOGITS

tadır

0.39

mbalikan

0.38

 itself

0.36

 gyhoeddwyd

0.35

RectangleBorder

0.35

XmlAccessorType

0.35

 stesso

0.35

abstractmethod

0.34

stdafx

0.33

 Aufstieg

0.33

Activations Density 0.055%

patterns related to numerical data and codes

No Comments

No Known Activations