INDEX

Explanations

mentions of abilities, actions, or states described with 'be' or 'have'

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

 Bars

-0.67

afia

-0.62

 promise

-0.62

 Genius

-0.61

 Might

-0.60

ossus

-0.59

 Extension

-0.59

 Whats

-0.58

 srfAttach

-0.58

 Mans

-0.57

POSITIVE LOGITS

 able

1.22

 rewarded

1.04

 judged

0.99

 replaced

0.98

heading

0.98

fall

0.95

ige

0.95

fitting

0.95

 phased

0.94

 punished

0.94

Activations Density 0.407%

mentions of abilities, actions, or states described with 'be' or 'have'

No Comments

No Known Activations