INDEX

Explanations

words related to specific languages

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ļ½

-0.67

 congr

-0.65

²¾

-0.63

 Recomm

-0.61

-+-+

-0.61

 confir

-0.61

OTOS

-0.61

 Bened

-0.60

ĳå£«

-0.60

 evacuation

-0.60

POSITIVE LOGITS

iland

1.09

ibaba

1.01

bucks

0.93

wear

0.90

kee

0.86

grass

0.81

kees

0.81

lde

0.80

erers

0.77

glers

0.77

Activations Density 0.144%

words related to specific languages

No Comments

No Known Activations