INDEX

Explanations

phrases related to people and their actions within news articles

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

 Guys

-0.65

 VIDEOS

-0.64

 Bound

-0.61

 Anything

-0.60

Georg

-0.59

IND

-0.58

 Meaning

-0.58

 economical

-0.56

UT

-0.56

SE

-0.56

POSITIVE LOGITS

 oversaw

1.18

 specializes

1.15

 owns

1.11

 oversees

1.08

 resided

1.08

 preceded

1.01

 attends

1.01

 specialize

1.00

 resides

1.00

 presided

0.99

Activations Density 0.584%

phrases related to people and their actions within news articles

No Comments

No Known Activations