INDEX

Explanations

phrases indicating assessment or evaluation processes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 21-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_21/width_16k/average_l0_129

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.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

 "-";

-0.43

 useAuth

-0.41

 FACULTY

-0.38

 @"";

-0.38

 Destination

-0.37

fahan

-0.37

 '-';

-0.36

])]

-0.35

TheReal

-0.35

?>/

-0.35

POSITIVE LOGITS

 help

0.65

 ensure

0.61

 مشين

0.56

OGND

0.56

AddTagHelper

0.54

 prevent

0.52

 щоб

0.52

 protect

0.51

 give

0.51

 CreateTagHelper

0.51

Activations Density 0.434%

phrases indicating assessment or evaluation processes

No Comments

No Known Activations