INDEX

Explanations

action words related to procedures or operations

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

tre

-0.66

wives

-0.64

conn

-0.64

voy

-0.64

zh

-0.62

rez

-0.60

sw

-0.60

rone

-0.60

wake

-0.59

tone

-0.59

POSITIVE LOGITS

ometimes

1.20

hift

0.83

 alot

0.75

paces

0.75

ilver

0.73

Tuc

0.71

omething

0.70

ynthesis

0.66

psey

0.66

 yearly

0.66

Activations Density 2.168%

action words related to procedures or operations

No Comments

No Known Activations