INDEX

Explanations

sentences that emphasize unity and common ground among individuals, often using the context of discussing differences in background, beliefs, or preferences

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

umbnail

-0.68

ociate

-0.65

luaj

-0.64

iling

-0.63

ertodd

-0.62

izons

-0.59

rongh

-0.59

ukong

-0.59

nel

-0.58

POSITIVE LOGITS

 raining

1.27

 unclear

1.26

 impossible

1.19

 imperative

1.14

 easier

1.11

 ironic

1.07

 doubtful

1.05

 advisable

1.05

 easy

1.04

 conceivable

1.03

Activations Density 3.005%

sentences that emphasize unity and common ground among individuals, often using the context of discussing differences in background, beliefs, or preferences

No Comments

No Known Activations