INDEX

Explanations

the word "one" surrounded by different context and phrases

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

ooks

-0.83

ories

-0.75

inders

-0.74

osponsors

-0.71

hips

-0.71

respective

-0.67

actions

-0.65

ourn

-0.63

emies

-0.62

folk

-0.61

POSITIVE LOGITS

 hundred

1.20

 Hundred

1.14

 thousand

1.01

 dimensional

0.89

 thing

0.88

 Thousand

0.85

 sided

0.85

 month

0.85

 hour

0.82

 Piece

0.82

Activations Density 0.389%

the word "one" surrounded by different context and phrases

No Comments

No Known Activations