INDEX

Explanations

punctuation marks, particularly periods, which may indicate the end of sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 5-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L5R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.5.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ÐµÑģÑĤÐ¸

-0.15

ertoire

-0.14

ãĥ³ãĥĨãĤ£

-0.13

.getUserName

-0.13

 Guerr

-0.13

Messages

-0.13

OrNull

-0.13

Ð¶ÐµÐ½

-0.13

sti

-0.13

zas

-0.13

POSITIVE LOGITS

alyzed

0.14

lda

0.14

amac

0.14

ctal

0.14

beros

0.14

robat

0.13

-basket

0.13

activex

0.13

tainment

0.13

ispiel

0.13

Activations Density 0.091%

punctuation marks, particularly periods, which may indicate the end of sentences

No Comments

No Known Activations