INDEX

Explanations

mentions of specific names

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

dayName

-0.72

haps

-0.70

 satell

-0.69

 jewels

-0.68

pmwiki

-0.67

fecture

-0.67

cele

-0.66

icted

-0.66

quiet

-0.66

 incor

-0.65

POSITIVE LOGITS

 Walsh

1.19

 McInt

1.19

 Hansen

1.16

 Snyder

1.15

 Tucker

1.12

 Allen

1.12

 Elliott

1.12

 Matthews

1.11

 Carter

1.11

 Smith

1.11

Activations Density 0.814%

mentions of specific names

No Comments

No Known Activations