INDEX

Explanations

terms related to variety and differences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

OIL

-0.82

kamp

-0.82

ovember

-0.81

ERO

-0.75

IER

-0.74

maxwell

-0.74

terday

-0.72

BALL

-0.71

emi

-0.71

{{

-0.71

POSITIVE LOGITS

 facets

1.32

 phases

1.17

 aspects

1.13

 kinds

1.09

 branches

1.06

 possibilities

1.05

 factions

1.02

 types

1.01

 iterations

1.00

 components

1.00

Activations Density 0.030%

terms related to variety and differences

No Comments

No Known Activations