INDEX

Explanations

instances of the word "so" used to emphasize or initiate a statement

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L1R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.1.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

acio

-0.16

adge

-0.16

contri

-0.15

 handler

-0.15

Syn

-0.14

locale

-0.13

criptors

-0.13

Inn

-0.13

olley

-0.13

.Hand

-0.13

POSITIVE LOGITS

ãģĵãĤį

0.16

.lbl

0.15

gue

0.14

åĬł

0.14

ÑĥÑĤÑĮ

0.13

iddles

0.13

ires

0.13

ething

0.13

:border

0.13

ãĤģãĤĭ

0.13

Activations Density 0.025%

instances of the word "so" used to emphasize or initiate a statement

No Comments

No Known Activations