INDEX

Explanations

references to images and image-related metadata

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

kit

-0.15

keit

-0.15

 circumstances

-0.14

 Pound

-0.14

asing

-0.14

ella

-0.14

ebo

-0.13

late

-0.13

 bust

-0.13

ambi

-0.13

POSITIVE LOGITS

 Resolution

0.20

Resolution

0.19

resolution

0.18

 resolution

0.18

 Dimension

0.18

 dimension

0.17

.resolution

0.16

Dimension

0.16

avin

0.16

aves

0.16

Activations Density 0.005%

references to images and image-related metadata

No Comments

No Known Activations