INDEX

Explanations

Verbs related to physical actions and processes or scientific and technical terms

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ļéĨĴ

-0.76

hound

-0.72

hower

-0.71

 deaf

-0.67

 friendly

-0.66

 "$:/

-0.66

EStream

-0.65

é¾įåĸļå£«

-0.63

 carriers

-0.61

gee

-0.60

POSITIVE LOGITS

rict

1.28

uffed

1.24

upid

1.23

uart

1.19

ructure

1.16

itched

1.16

alker

1.15

rikes

1.15

amped

1.12

ocking

1.10

Activations Density 0.028%

Verbs related to physical actions and processes or scientific and technical terms

No Comments

No Known Activations