INDEX

Explanations

names of places and people

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

PDATE

-0.71

STD

-0.70

Spot

-0.68

AUD

-0.68

 crunch

-0.66

IOR

-0.66

NPR

-0.65

âĸ¬

-0.65

ĸļ

-0.64

bilt

-0.64

POSITIVE LOGITS

phia

1.09

hire

0.85

ensis

0.81

 Parish

0.80

oulos

0.77

utics

0.76

 Palace

0.76

icz

0.75

tein

0.75

rosis

0.75

Activations Density 1.790%

names of places and people

No Comments

No Known Activations