INDEX

Explanations

technical questions related to programming or computer science concepts.

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

metadata related to programming, technical questions, and solutions.

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

the beginning of text passages or documents. The neuron shows very high activation for "<bos>" tokens, which typically indicate the start of a new input sequence or document.

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

 sauvages

-0.52

hima

-0.51

ajah

-0.50

 fallu

-0.49

او

-0.48

 proprement

-0.46

 moindre

-0.45

 ciudadana

-0.45

Rüyada

-0.45

에서는

-0.45

POSITIVE LOGITS

bitField

0.90

 AssemblyVersion

0.79

Personendaten

0.76

 unknownFields

0.75

UnknownFieldSet

0.73

 Roskov

0.70

awtextra

0.70

ThroughAttribute

0.70

 الرياضيه

0.69

MessageTagHelper

0.68

Activations Density 0.104%

technical questions related to programming or computer science concepts.

metadata related to programming, technical questions, and solutions.

the beginning of text passages or documents. The neuron shows very high activation for "<bos>" tokens, which typically indicate the start of a new input sequence or document.

No Comments

No Known Activations