INDEX

Explanations

unique character sequences or symbols that may not correspond to typical textual content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 31-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L31R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.31.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

ilim

-0.14

Ã«l

-0.14

elan

-0.13

Ð»ÐµÑĩ

-0.13

Bas

-0.13

 useClass

-0.13

olog

-0.13

bourg

-0.13

eres

-0.13

oke

-0.13

POSITIVE LOGITS

 back

0.22

=back

0.20

back

0.20

(back

0.19

Back

0.19

 whatever

0.19

-back

0.19

 Back

0.18

/back

0.18

 backed

0.18

Activations Density 0.011%

unique character sequences or symbols that may not correspond to typical textual content

No Comments

No Known Activations