INDEX

Explanations

phrases that indicate contrast or exceptions in a discussion or argument

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

but and except clauses

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 voidaan

-0.32

 auguri

-0.32

 cię

-0.32

 serdecznie

-0.32

akses

-0.32

 debió

-0.31

 advises

-0.30

 Geister

-0.30

rowad

-0.30

Status

-0.30

POSITIVE LOGITS

OGND

0.69

ngrx

0.64

DockStyle

0.64

 protoimpl

0.64

SharedCtor

0.62

BufferException

0.60

yntaxException

0.60

EDEFAULT

0.60

AndEndTag

0.60

ValueStyle

0.60

Activations Density 0.050%

phrases that indicate contrast or exceptions in a discussion or argument

but and except clauses

No Comments

No Known Activations