INDEX

Explanations

words indicating ongoing actions, conditions, or expectations

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

IActionResult

-0.47

reszcie

-0.45

Izvori

-0.44

 gangen

-0.44

openConnection

-0.43

 coroa

-0.42

engagent

-0.42

 granic

-0.42

RequestOptions

-0.41

 Reiche

-0.41

POSITIVE LOGITS

STILL

0.75

Still

0.75

still

0.70

 Still

0.65

 STILL

0.64

 ſtill

0.64

 snippetHide

0.62

 still

0.60

 comunque

0.60

 ftill

0.60

Activations Density 0.039%

words indicating ongoing actions, conditions, or expectations

No Comments

No Known Activations