INDEX

Explanations

dates, numbers, locations, and rankings within texts

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

 glim

-0.79

othe

-0.73

 needle

-0.66

 shar

-0.66

 polic

-0.63

ople

-0.63

 serv

-0.63

 corrid

-0.62

 ranc

-0.62

omorph

-0.60

POSITIVE LOGITS

1.75

1.73

1.72

1.71

Activations Density 0.084%

dates, numbers, locations, and rankings within texts

No Comments

No Known Activations