INDEX

Explanations

reasons or explanations in a text

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

semble

-0.74

ibaba

-0.67

ymph

-0.66

 Roller

-0.64

ault

-0.63

chron

-0.63

 Carbuncle

-0.61

rop

-0.61

izen

-0.60

 transm

-0.60

POSITIVE LOGITS

why

1.37

why

1.12

WHY

1.09

abl

1.02

Why

0.84

Why

0.82

 justifying

0.80

Origin

0.77

 cele

0.76

 rationale

0.73

Activations Density 1.551%

reasons or explanations in a text

No Comments

No Known Activations