INDEX

Explanations

information related to research studies and polling data

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

inventoryQuantity

-0.77

 brakes

-0.76

 retaliate

-0.75

 shutting

-0.75

 hurting

-0.75

 tant

-0.73

mare

-0.72

danger

-0.72

Shut

-0.69

 retaliation

-0.67

POSITIVE LOGITS

 randomized

1.34

 dataset

1.30

 statistical

1.20

 analysed

1.18

 randomly

1.17

 analyzed

1.17

 chronological

1.14

 datasets

1.14

 sample

1.13

 sampled

1.11

Activations Density 5.415%

information related to research studies and polling data

No Comments

No Known Activations