INDEX

Explanations

statements where someone is not directly revealing specific information

oai_token-act-pair · gpt-3.5-turbo

instances of uncertainty or reluctance to disclose information

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itton

-0.79

tic

-0.70

naissance

-0.66

itans

-0.65

ngth

-0.62

ixture

-0.61

aturated

-0.61

ixtures

-0.61

jam

-0.61

ickle

-0.60

POSITIVE LOGITS

 specifics

1.46

 anything

1.13

nor

1.07

any

1.00

 anymore

1.00

 definitively

0.98

 exactly

0.94

 specific

0.90

 publicly

0.89

 whether

0.89

Activations Density 0.140%

statements where someone is not directly revealing specific information

instances of uncertainty or reluctance to disclose information

No Comments

No Known Activations