INDEX

Explanations

objects and actions related to manual tasks and tools

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

 knife

-0.18

 knives

-0.17

needle

-0.17

 needle

-0.16

 scissors

-0.16

 needles

-0.15

åĪĢ

-0.15

 Knife

-0.15

maze

-0.15

 DeepCopy

-0.14

POSITIVE LOGITS

 hammer

0.40

 Hammer

0.34

hammer

0.30

Ham

0.26

 club

0.26

.ham

0.26

 clubs

0.25

ammers

0.24

Club

0.24

club

0.22

Activations Density 0.067%

objects and actions related to manual tasks and tools

No Comments

No Known Activations