INDEX

Explanations

terms related to training, education, and healthcare benefits

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

UGH

-0.16

ough

-0.16

æ³Ĭ

-0.15

paralle

-0.14

 Apprec

-0.14

 independ

-0.14

ctic

-0.14

orna

-0.13

inding

-0.13

oyal

-0.13

POSITIVE LOGITS

ÄĽj

0.15

Ø§ÙħØ§

0.14

 MÃ¶

0.14

istÃ©

0.14

inel

0.14

yb

0.14

ropriate

0.14

_GLOBAL

0.14

 Entr

0.14

sir

0.13

Activations Density 0.032%

terms related to training, education, and healthcare benefits

No Comments

No Known Activations