INDEX

Explanations

phrases related to giving instructions or making authoritative commands

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

pmwiki

-0.86

nown

-0.76

cit

-0.75

attery

-0.74

sonian

-0.70

ãĤ§

-0.70

doi

-0.70

uded

-0.69

uggets

-0.69

 Echoes

-0.69

POSITIVE LOGITS

 deported

0.79

 execution

0.76

 executions

0.75

 extermin

0.73

FontSize

0.71

 demolition

0.71

 shutdown

0.70

 attendance

0.70

 inspections

0.70

lies

0.70

Activations Density 0.040%

phrases related to giving instructions or making authoritative commands

No Comments

No Known Activations