INDEX

Explanations

mentions of specific names and terms

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

-0.66

ext

-0.57

 entrants

-0.56

 blazing

-0.55

FIG

-0.55

 ISBN

-0.53

 rabbit

-0.53

 litter

-0.53

 Camb

-0.52

âĵĺ

-0.52

POSITIVE LOGITS

heim

0.94

otropic

0.79

mentioned

0.78

wered

0.77

 Pradesh

0.77

sembly

0.73

ndra

0.70

urnal

0.70

erman

0.69

asia

0.68

Activations Density 2.090%

mentions of specific names and terms

No Comments

No Known Activations