INDEX

Explanations

instances of "first" and related concepts indicating initial occurrences or significance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 7-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L7R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.7.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.17

essim

-0.16

FindObject

-0.14

Naw

-0.14

 camp

-0.14

DET

-0.14

izont

-0.14

çŃĭ

-0.14

 force

-0.14

rem

-0.13

POSITIVE LOGITS

enburg

0.19

neau

0.17

aid

0.17

Ð°ÑģÑĤÐ¸

0.16

ÑĢÐ°Ð²Ð¸

0.15

į¨

0.15

aison

0.15

ough

0.14

eny

0.14

.getResponse

0.14

Activations Density 0.118%

instances of "first" and related concepts indicating initial occurrences or significance

No Comments

No Known Activations