INDEX

Explanations

geographical locations

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

yip

-0.86

verty

-0.65

HDL

-0.60

agg

-0.58

 Mahjong

-0.56

ashtra

-0.56

 Dragonbound

-0.55

perty

-0.55

grain

-0.54

olini

-0.54

POSITIVE LOGITS

alion

0.83

ciating

0.79

liction

0.73

inition

0.71

igm

0.69

orable

0.67

agram

0.67

amation

0.65

hyde

0.65

pora

0.64

Activations Density 0.056%

geographical locations

No Comments

No Known Activations