INDEX

Explanations

references to numbers or quantities, particularly focusing on the terms "two" and "four"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 16-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L16R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.16.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

 importantly

-0.17

{{{

-0.16

enna

-0.16

ena

-0.15

ello

-0.15

yna

-0.15

 fewer

-0.14

 Hutchinson

-0.14

 ears

-0.14

allas

-0.14

POSITIVE LOGITS

oose

0.19

eker

0.16

HITE

0.16

latest

0.15

 remaining

0.15

egie

0.15

é¨

0.15

remaining

0.14

 available

0.14

actable

0.14

Activations Density 0.177%

references to numbers or quantities, particularly focusing on the terms "two" and "four"

No Comments

No Known Activations