INDEX

Explanations

mentions of events, people, and specific 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_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

nect

-0.80

 seeker

-0.78

ĺħ

-0.76

 obser

-0.69

baugh

-0.67

ãģĨ

-0.67

 planner

-0.64

cale

-0.62

 seekers

-0.62

missions

-0.61

POSITIVE LOGITS

ting

1.31

zeb

1.07

tery

1.05

hent

1.03

tering

1.01

cher

0.98

ter

0.95

rodu

0.93

rans

0.91

arget

0.89

Activations Density 1.160%

mentions of events, people, and specific locations

No Comments

No Known Activations