INDEX

Explanations

references to time-related phrases or sequences

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

muz

-0.16

umo

-0.15

borough

-0.15

uco

-0.14

ish

-0.14

rahim

-0.14

ify

-0.14

seau

-0.14

ffen

-0.14

.Mapping

-0.13

POSITIVE LOGITS

ãĢħ

0.25

 nhau

0.15

à¹Ĩ

0.15

 à¹Ĩ

0.15

ÑĢÐ¾ÑĪ

0.14

/from

0.14

 merk

0.13

 another

0.13

 hour

0.13

íĦ

0.13

Activations Density 0.036%

references to time-related phrases or sequences

No Comments

No Known Activations