INDEX

Explanations

references to specific dates or historical events

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

inational

-0.68

entric

-0.66

Osc

-0.63

viz

-0.59

 Blues

-0.58

aukee

-0.57

ciation

-0.56

risome

-0.56

 Newly

-0.56

ISION

-0.56

POSITIVE LOGITS

stab

1.21

packs

1.20

dated

1.17

GROUND

1.15

tracking

1.11

packing

1.10

side

1.07

lash

1.07

doors

1.07

wards

1.05

Activations Density 2.241%

references to specific dates or historical events

No Comments

No Known Activations