INDEX

Explanations

specific structured information or references to detailed items in various contexts

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

цездатний

-0.77

IsContent

-0.69

 виправивши

-0.68

 nakalista

-0.66

rungsseite

-0.66

 المعيارى

-0.63

 Comprometido

-0.62

 '\\;'

-0.60

 שוליים

-0.59

FormTagHelper

-0.59

POSITIVE LOGITS

 originally

0.33

sty

0.32

 John

0.32

aer

0.31

ClientSize

0.30

 internally

0.29

 sealed

0.29

 Centi

0.29

 hela

0.29

 expected

0.29

Activations Density 0.115%

specific structured information or references to detailed items in various contexts

No Comments

No Known Activations