INDEX

Explanations

terms related to negativity or pejorative expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 4-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L4R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.4.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

hes

-0.20

reload

-0.18

ships

-0.17

ship

-0.17

land

-0.15

lands

-0.15

viso

-0.15

so

-0.14

nom

-0.14

li

-0.14

POSITIVE LOGITS

croft

0.17

-pe

0.16

alties

0.15

atures

0.15

rine

0.15

ÑĢÑĥÑĪ

0.15

aucoup

0.14

bean

0.14

ijke

0.14

OfString

0.14

Activations Density 0.019%

terms related to negativity or pejorative expressions

No Comments

No Known Activations