INDEX

Explanations

statements and reports in interviews, emails, and statements

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

atics

-0.75

 needles

-0.69

 sooner

-0.65

 proven

-0.64

 broom

-0.64

abiding

-0.64

aters

-0.62

Mods

-0.62

iors

-0.62

 wherever

-0.61

POSITIVE LOGITS

 HuffPost

0.97

 remarks

0.95

 Thursday

0.94

 Wednesday

0.94

 briefing

0.93

 Tuesday

0.93

 Friday

0.92

 POLITICO

0.89

 announcing

0.89

 keynote

0.88

Activations Density 4.700%

statements and reports in interviews, emails, and statements

No Comments

No Known Activations