INDEX

Explanations

words related to thoughts, beliefs, expectations, or fears

oai_token-act-pair · gpt-3.5-turbo

expressions of fear, belief, and expectation regarding uncertain outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

arta

-0.74

aday

-0.69

ety

-0.69

vasive

-0.69

versions

-0.69

tha

-0.69

ighters

-0.68

washer

-0.68

oven

-0.67

por

-0.67

POSITIVE LOGITS

 originally

0.70

 initially

0.68

lessly

0.66

 beforehand

0.65

 unsuccessfully

0.63

 innocence

0.61

 errone

0.61

auga

0.61

 underestimated

0.61

hes

0.61

Activations Density 0.141%

words related to thoughts, beliefs, expectations, or fears

expressions of fear, belief, and expectation regarding uncertain outcomes

No Comments

No Known Activations