INDEX

Explanations

the word "over" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_0/width_65k/average_l0_72

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.hook_mlp_out

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

over

-1.75

overs

-0.93

able

-0.93

\{\\

-0.85

out

-0.84

ology

-0.82

about

-0.80

OVER

-0.79

above

-0.74

away

-0.71

POSITIVE LOGITS

ing

0.87

ING

0.63

 CreateTagHelper

0.63

nements

0.60

---*/

0.59

ляр

0.57

်

0.56

Nowhere

0.56

н

0.55

нг

0.54

Activations Density 0.025%

the word "over" in various contexts

No Comments

No Known Activations