INDEX

Explanations

time-related expressions and references

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

emort

-0.82

ngth

-0.78

negie

-0.76

ACTED

-0.74

obal

-0.72

egal

-0.72

essee

-0.71

icide

-0.71

hett

-0.70

bourg

-0.70

POSITIVE LOGITS

dream

1.68

care

1.11

lights

0.99

break

0.96

light

0.87

trip

0.86

long

0.85

motion

0.83

life

0.81

lighting

0.81

Activations Density 0.436%

time-related expressions and references

No Comments

No Known Activations