INDEX

Explanations

conjunctions and prepositions connected with law and political contexts

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

late

-0.87

raq

-0.84

ãĥī

-0.74

geist

-0.72

mun

-0.72

iers

-0.72

cow

-0.70

mers

-0.68

lish

-0.66

estate

-0.66

POSITIVE LOGITS

 knowing

1.03

 regard

1.02

 mentioning

0.99

 hesitation

0.97

 interruption

0.95

 sacrificing

0.94

 exception

0.93

 necessarily

0.93

 specifying

0.90

 recourse

0.89

Activations Density 0.872%

conjunctions and prepositions connected with law and political contexts

No Comments

No Known Activations