INDEX

Explanations

terms related to the concept of allowing or permitting actions or features

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 30-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L30R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.30.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

edImage

-0.15

bower

-0.15

edList

-0.13

atories

-0.13

bows

-0.13

culus

-0.13

comma

-0.12

fried

-0.12

esian

-0.12

/tools

-0.12

POSITIVE LOGITS

0.89

te

0.78

ts

0.72

td

0.66

ty

0.66

tes

0.66

ta

0.66

ti

0.65

ting

0.64

ten

0.64

Activations Density 0.427%

terms related to the concept of allowing or permitting actions or features

No Comments

No Known Activations