INDEX

Explanations

titles or headings of reports or sections

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

quo

-0.65

 suppressed

-0.65

 dece

-0.64

 vain

-0.63

etheless

-0.63

 slowing

-0.63

 infiltration

-0.61

 diver

-0.61

 discharge

-0.60

 srfAttach

-0.60

POSITIVE LOGITS

atson

0.88

oran

0.85

itialized

0.83

chin

0.83

edes

0.82

zac

0.81

vent

0.79

mans

0.77

ussie

0.76

slaught

0.76

Activations Density 1.215%

titles or headings of reports or sections

No Comments

No Known Activations