INDEX

Explanations

concepts related to reasoning and evaluation

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

 pleaſure

-0.40

 faſt

-0.39

 itſelf

-0.39

 juſ

-0.36

Modifiers

-0.36

IFICATE

-0.36

 toek

-0.35

来得

-0.35

 поскольку

-0.35

 particuliers

-0.35

POSITIVE LOGITS

0.58

 typelib

0.56

 BorderSide

0.56

VolleyError

0.52

 Infórmanos

0.50

ագրություններ

0.48

Espèce

0.48

 ProtoMessage

0.47

ViewImports

0.47

.*")]

0.47

Activations Density 0.071%

concepts related to reasoning and evaluation

No Comments

No Known Activations