INDEX

Explanations

sentences that contain high activation words and phrases indicating significant actions or concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

oin

-0.17

ucha

-0.16

Builder

-0.16

Ã¡f

-0.15

quare

-0.15

tro

-0.15

assa

-0.15

uja

-0.14

oui

-0.14

allah

-0.14

POSITIVE LOGITS

.@

0.18

 Sources

0.16

 Ahead

0.15

ervas

0.15

 sources

0.15

Sources

0.14

_charset

0.14

Meanwhile

0.14

artz

0.14

_indx

0.14

Activations Density 0.007%

sentences that contain high activation words and phrases indicating significant actions or concepts

No Comments

No Known Activations