INDEX

Explanations

superlatives or comparative adjectives and nouns related to entertainment or technology

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

skirts

-0.82

kas

-0.73

bold

-0.71

ttes

-0.68

mun

-0.68

attle

-0.65

space

-0.65

rahim

-0.65

chy

-0.64

md

-0.64

POSITIVE LOGITS

 possible

1.11

 imaginable

1.04

 conceivable

1.02

seller

0.97

 ever

0.92

 practicable

0.91

hest

0.86

iary

0.85

 exponent

0.82

 predictor

0.78

Activations Density 1.073%

superlatives or comparative adjectives and nouns related to entertainment or technology

No Comments

No Known Activations