INDEX

Explanations

words related to political figures and activities, including comparisons, roles, and events

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

calling

-0.64

 wastes

-0.61

Got

-0.59

ain

-0.57

need

-0.57

pict

-0.57

 rolled

-0.56

 packets

-0.55

 folded

-0.55

 Donation

-0.55

POSITIVE LOGITS

ggles

1.40

 ensure

1.34

ilet

1.33

 maximize

1.28

 compensate

1.28

 commemorate

1.28

 avoid

1.26

 satisfy

1.26

 minimize

1.24

 accommodate

1.22

Activations Density 1.727%

words related to political figures and activities, including comparisons, roles, and events

No Comments

No Known Activations