INDEX

Explanations

various instances of ''partly'' phrases and causal relationships

oai_token-act-pair · gpt-3.5-turbo

phrases indicating causation or reason

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

urch

-0.74

apest

-0.69

thens

-0.69

mons

-0.65

ç«

-0.63

thening

-0.63

inals

-0.61

 unbeaten

-0.61

Express

-0.61

ature

-0.60

POSITIVE LOGITS

 attributable

1.23

 attributed

1.12

due

1.03

 stemming

1.00

 owing

0.93

 generational

0.89

due

0.88

 stemmed

0.88

 related

0.85

 genetics

0.84

Activations Density 0.291%

various instances of ''partly'' phrases and causal relationships

phrases indicating causation or reason

No Comments

No Known Activations