INDEX

Explanations

phrases related to news articles reporting events, such as locations, times, people involved, and activities happening

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

ictionary

-0.78

govtrack

-0.71

ledge

-0.70

liam

-0.68

Layer

-0.67

linux

-0.67

ingly

-0.67

ternal

-0.66

heit

-0.65

 alike

-0.65

POSITIVE LOGITS

 latter

1.38

 latest

1.27

 aforementioned

1.16

 situation

1.15

 incident

1.14

 offending

1.09

 ordeal

1.08

 outcome

1.06

 same

1.05

 remainder

1.05

Activations Density 2.003%

phrases related to news articles reporting events, such as locations, times, people involved, and activities happening

No Comments

No Known Activations