INDEX

Explanations

references to secret or hidden information or locations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stand

-0.70

boxes

-0.68

tops

-0.68

board

-0.67

beard

-0.61

strom

-0.60

 Taiwanese

-0.60

winner

-0.59

Fram

-0.59

verbs

-0.59

POSITIVE LOGITS

ribed

1.16

ript

1.04

apes

1.01

aped

0.98

ence

0.95

ribing

0.91

utions

0.89

opes

0.88

apers

0.88

uers

0.87

Activations Density 0.016%

references to secret or hidden information or locations

No Comments

No Known Activations