INDEX

Explanations

mentions of specific movie titles and characters from a superhero franchise

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

ickr

-0.85

jri

-0.84

mediate

-0.83

unicip

-0.82

elong

-0.82

tu

-0.81

 administr

-0.80

 benches

-0.79

housing

-0.79

prus

-0.77

POSITIVE LOGITS

 trilogy

1.51

 Chronicles

1.48

 sequel

1.46

 Trilogy

1.44

 sequels

1.42

 novels

1.28

 starring

1.26

 Adventures

1.26

 Dracula

1.25

 Episode

1.23

Activations Density 4.992%

mentions of specific movie titles and characters from a superhero franchise

No Comments

No Known Activations