INDEX

Explanations

occurrences of the phrase "what [someone] said" or "what [someone] mentioned"

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

Not in Any Lists

No Comments

Negative Logits

ãĥĺ

-0.87

¨

-0.79

alties

-0.79

²¾

-0.77

lees

-0.75

Ħ¢

-0.75

aneers

-0.72

ãĥł

-0.71

abilities

-0.70

flies

-0.69

POSITIVE LOGITS

 excerpt

1.06

 rundown

1.03

 synopsis

1.00

 breakdown

0.91

 overview

0.89

 summary

0.83

 nutshell

0.83

 kicker

0.82

 cached

0.79

 redacted

0.78

Activations Density 0.122%

occurrences of the phrase "what [someone] said" or "what [someone] mentioned"

No Comments

No Known Activations