INDEX

Explanations

the word "be" used in various contexts

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

PsyNetMessage

-0.69

 fray

-0.67

 srfAttach

-0.67

 Times

-0.65

 Prescott

-0.63

 Taxes

-0.62

 Wick

-0.60

 Stain

-0.60

Wid

-0.60

 residue

-0.59

POSITIVE LOGITS

 able

1.41

 judged

1.11

 aware

1.03

hemoth

1.02

 treated

1.00

heading

0.99

 replaced

0.98

 evaluated

0.98

 rewarded

0.98

 punished

0.97

Activations Density 1.030%

the word "be" used in various contexts

No Comments

No Known Activations