INDEX

Explanations

negative contractions and their associated contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.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

ÙģØ´

-0.14

Jur

-0.14

ardin

-0.14

sworth

-0.14

 Seks

-0.14

erten

-0.14

subst

-0.14

udu

-0.14

foy

-0.13

POSITIVE LOGITS

.LayoutStyle

0.15

aeper

0.14

ÙĴØ±

0.14

³

0.14

 conviction

0.14

chan

0.14

 Jaune

0.14

endon

0.14

HI

0.13

ForRow

0.13

Activations Density 0.026%

negative contractions and their associated contexts

No Comments

No Known Activations