INDEX

Explanations

phrases related to specific actions that need to be taken or accomplished

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

Written

-0.64

dozen

-0.63

 Tent

-0.59

 Cosponsors

-0.59

written

-0.58

 Methodist

-0.58

Typ

-0.57

 Gone

-0.56

wrote

-0.56

 Verse

-0.55

POSITIVE LOGITS

 maximize

1.13

 ensure

1.04

 insure

0.99

 avoid

0.96

 facilitate

0.96

 assure

0.92

 minimize

0.90

 optimize

0.89

 prevent

0.89

 qualify

0.89

Activations Density 2.922%

phrases related to specific actions that need to be taken or accomplished

No Comments

No Known Activations