INDEX

Explanations

sentence transitions or contrasts, particularly words like "but", "however", and phrases that introduce new information or qualifications to previous statements.

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

discussions about performance metrics and technical specifications.

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

technical details and code information in a document.

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

Text after punctuation (., ?, or comma)

np_acts-logits-general · gemini-2.0-flash

again

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

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

WebServlet

-0.66

 Derfor

-0.64

 BoxFit

-0.58

çadas

-0.57

,:);

-0.57

phic

-0.55

]^{-

-0.54

ïsme

-0.51

xec

-0.51

 marbre

-0.50

POSITIVE LOGITS

 again

3.62

Again

3.51

 Again

3.44

again

3.44

 AGAIN

2.62

 novamente

2.49

 nuevamente

2.39

 opět

2.36

 опять

2.35

こちらも

2.09

Activations Density 0.461%