INDEX

Explanations

content related to politics, policy, and healthcare discussions

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 shroud

-0.67

 Samoa

-0.67

 Marketable

-0.66

 habit

-0.61

 scatter

-0.60

ettings

-0.60

 Outer

-0.60

 Cellular

-0.59

OTOS

-0.59

Sic

-0.58

POSITIVE LOGITS

º

0.99

¡

0.90

£

0.89

ľ

0.88

especially

0.86

said

0.86

¬

0.85

ķ

0.84

yet

0.84

Ń

0.82

Activations Density 1.343%

content related to politics, policy, and healthcare discussions

No Comments

No Known Activations