INDEX

Explanations

phrases and terms related to instructions and instruction manuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

ogia

-0.59

rrggbb

-0.57

("")){

-0.56

└──

-0.55

 Hickey

-0.55

 مشين

-0.54

stateProvider

-0.54

찮

-0.53

thâu

-0.53

оне

-0.52

POSITIVE LOGITS

 instructions

0.78

 propOrder

0.68

 instru

0.66

 instructed

0.66

 instruction

0.64

 INSTRUCTION

0.64

INSTRUCTION

0.64

CppMethod

0.63

Instru

0.62

 Instructions

0.62

Activations Density 0.123%

phrases and terms related to instructions and instruction manuals

No Comments

No Known Activations