INDEX

Explanations

occurrences of the word "each" followed by a number, emphasizing individual instances or quantities

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

haven

-0.73

tops

-0.72

rovers

-0.72

stones

-0.71

children

-0.68

abs

-0.68

models

-0.65

marks

-0.63

bows

-0.62

letters

-0.62

POSITIVE LOGITS

 successive

1.24

 iteration

1.02

 individual

0.97

 individually

0.95

 respective

0.91

 succeeding

0.84

 element

0.83

 participant

0.81

 ounce

0.81

 side

0.81

Activations Density 0.299%

occurrences of the word "each" followed by a number, emphasizing individual instances or quantities

No Comments

No Known Activations