INDEX

Explanations

instances of the prefix "incon," indicating themes of incongruity or inconsistency

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 19-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L19R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.19.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arna

-0.16

Gle

-0.15

ascar

-0.14

tay

-0.14

itet

-0.14

Ð»Ð¾Ð²Ð°

-0.14

eled

-0.14

endum

-0.14

iska

-0.14

Priv

-0.14

POSITIVE LOGITS

 incom

0.15

gv

0.15

 permanent

0.14

alance

0.14

 Fighters

0.14

jac

0.14

æİī

0.14

/bus

0.14

 Colomb

0.14

ë¦¬ìķĦ

0.13

Activations Density 0.046%

instances of the prefix "incon," indicating themes of incongruity or inconsistency

No Comments

No Known Activations