INDEX

Explanations

references to seasons and episodes in TV shows

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

IFrame

Link

Negative Logits

uckland

-0.75

ught

-0.75

alty

-0.67

ifles

-0.62

insula

-0.60

issance

-0.59

anguages

-0.59

eport

-0.59

 clutch

-0.57

ibrary

-0.56

POSITIVE LOGITS

VII

1.25

 VIII

1.21

III

1.18

XII

1.14

1.12

1.09

IV

1.09

1.08

1.07

Activations Density 0.223%

references to seasons and episodes in TV shows

No Known Activations