INDEX

Explanations

language related to evaluating risk and making judgments.

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

words related to opinions and speculations.

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases or statements that express uncertainty, rumors, or contradictions to expectations. It tends to activate strongly for words and phrases like "rumors", "fears", "if", "wasn't true", "incorrect", "you'd be wrong", etc. This suggests the neuron is detecting language related to

oai_token-act-pair · claude-3-5-sonnet-20240620 Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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



-0.86

 nahilalakip

-0.76

 kasarigan

-0.73

SourceChecksum

-0.69

 Италијани

-0.69

 '\\;'

-0.68

Tikang

-0.66

expandindo

-0.65

 EconPapers

-0.63

principalColumn

-0.63

POSITIVE LOGITS

 truth

1.37

 correct

1.32

 accurate

1.31

 valid

1.30

 accuracy

1.27

 validity

1.27

 correctness

1.16

 true

1.13

 veracity

1.13

correct

1.11

Activations Density 0.481%

language related to evaluating risk and making judgments.

words related to opinions and speculations.

phrases or statements that express uncertainty, rumors, or contradictions to expectations. It tends to activate strongly for words and phrases like "rumors", "fears", "if", "wasn't true", "incorrect", "you'd be wrong", etc. This suggests the neuron is detecting language related to

No Comments

No Known Activations