INDEX

Explanations

text related to philosophical, political, or historical 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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 predec

-0.70

 buggy

-0.70

 scatter

-0.67

 stricken

-0.66

 lodging

-0.65

 shroud

-0.64

 clad

-0.64

 neglig

-0.63

 closest

-0.63

 decomp

-0.63

POSITIVE LOGITS

º

1.23

£

1.09

¹

1.07

¡

0.94

®

0.94

į

0.91

¬

0.91

»

0.90

Ĵ

0.89

Ń

0.88

Activations Density 0.236%

text related to philosophical, political, or historical contexts

No Comments

No Known Activations