INDEX

Explanations

analysis, interpretation, argument

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 narratives

-0.11

 conceptual

-0.11

 scenarios

-0.10

 elabor

-0.10

è¿°

-0.10

 repl

-0.09

 narrative

-0.09

 Narrative

-0.09

ÙĪØ´

-0.09

.writeValue

-0.09

POSITIVE LOGITS

 analysis

0.19

åĪĨæŀĲ

0.19

Analysis

0.17

 Analysis

0.17

 interpretation

0.17

analysis

0.17

-analysis

0.16

 Interpret

0.16

 interpret

0.15

_analysis

0.15

Activations Density 0.205%