INDEX

Explanations

words and suffixes related to medical or chemical terminology, especially those ending in specific patterns

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ivec

-0.15

ilda

-0.15

 External

-0.14

ä»¶

-0.14

ãĥ¬ãĥĵ

-0.14

bak

-0.14

istics

-0.14

ifo

-0.14

Kin

-0.13

iences

-0.13

POSITIVE LOGITS

sis

0.25

lation

0.25

nnen

0.24

ellow

0.23

ez

0.23

nn

0.23

tic

0.23

ea

0.23

eah

0.23

Activations Density 0.065%

words and suffixes related to medical or chemical terminology, especially those ending in specific patterns

No Comments

No Known Activations