2,932,904 events, 1,463,475 push events, 2,431,402 commit messages, 272,281,747 characters
Create 5. Do blowouts translate to lost viewers?
The vast majority of Super Bowls are close games. Makes sense. Both teams are likely to be deserving if they've made it this far. The closest game ever was when the Buffalo Bills lost to the New York Giants by 1 point in 1991, which was best remembered for Scott Norwood's last-second missed field goal attempt that went wide right, kicking off four Bills Super Bowl losses in a row. Poor Scott. The biggest point discrepancy ever was 45 points (!) where Hall of Famer Joe Montana's led the San Francisco 49ers to victory in 1990, one year before the closest game ever. I remember watching the Seahawks crush the Broncos by 35 points (43-8) in 2014, which was a boring experience in my opinion. The game was never really close. I'm pretty sure we changed the channel at the end of the third quarter. Let's combine our game data and TV to see if this is a universal phenomenon. Do large point differences translate to lost viewers? We can plot household share (average percentage of U.S. households with a TV in use that were watching for the entire broadcast) vs. point difference to find out.
Require prettier formatting again
Prior to 553122891 in August 2019, this repository required all code to be formatted with the tool Prettier. We decided to remove this requirement; @abernix included a compelling explanation in that commit's message.
The state of prettier in the repository has been somewhat mixed recently. I'm currently the lead maintainer of Apollo Server and I've kept all the files I work on most actively (including most of apollo-server-core) prettier-clean when feasible. In my personal experience I've found that the ability to completely turn off any part of my brain that thinks about the details of formatting outweighs the drawbacks described in the above message. I've also found that many contributors do run prettier accidentally as part of their commits; telling people to revert changes (which often increase readability!) feels like as much of a waste of time as forcing folks who may have never encountered the quite popular Prettier tool to run it. We do use enforced Prettier on several of Apollo's internal projects and it does seem to work well.
As an experiment, we're going to try turning Prettier enforcement back on. If this proves to be a hurdle for contributors, we can revert it. I recognize there's some hubris in ignoring @abernix's well-reasoned plea from two years ago, and there's a very good chance that in a few months I'll have learned the hard way that his judgment two years ago was better than my judgment today.
Note that since we've upgraded to Prettier 2 since the last time we enforced Prettier, so as the comment in the old prettierrc suggests, we no longer need its requirePragma hack.
Benefits Of Couples Massage Tukwila
Your body releases the "love hormone oxytocin", which gives the massage a feeling of love and affection. This means that when you massage couples, Tukwila's love is in the air (and in the bloodstream). Increased appreciation is one of the main benefits of multiple massages. Many couples want to have a good time together during their hectic times, and multiple massage sessions can be a deliberate and deliberate reunion. Massage can also improve well-being and intimacy by releasing oxytocin, serotonin, and dopamine. According to experts, such statistics suggest that couples may have frequent massage sessions.
For more: https://sites.google.com/view/bluelotusspa/blog/benefits-of-couples-massage-tukwila
Couples Massage Tukwila
Your body releases the 'love hormone oxytocin,' which gives your massages a sense of love and affection. It implies that when you get a Couple Massage, Tukwila love is in the air (and the bloodstream). Increased appreciation is one of the main advantages of a few massages. Many couples want to spend quality time together in their hectic times, and a few massages may be deliberate and deliberate reconnection. Massage treatment may also help enhance the condition and vicinity by producing oxytocin’s, serotonin, and body dopamine. According to specialists, such statistics suggest that couples may have frequent massages.
For more: https://tananet.net/couple-massage-tukwila/
"11:05am. I went to bed a bit earlier, but I failed at moving the internal wake up clock. Only got up 10m ago and I feel groggy right now. Well, either way today is the programming day. I am not going to waste time waiting for my grogginess to pass. Instead, I am going to get started on the transformer and make my way from there.
I've made up my resolve. To break into the 4/5 realm where I can exert heavy pressure on the real world, I need energy based models, algos for them and neurochips. I am not going to force deep learning into the mold. It won't be able to handle composition of independently trained modules nor long term credit assignment.
Instead I'll aim to reach the pinnacle of 3/5. The current deep learning wave is here and I should just ride it out to its end. It is time to master it thoroughly.
I've spent way too much time chasing down empty leads. It is time to get things done.
11:10am. I know I say this often, but once I implement the SAU trick, there will be no way to exceed the current formulation very easily. Not without some complicated GAN prediction training. What I have in mind now is as easy as it will get. Low hanging fruit like SAU was truly a furtuitous encounter.
11:25am.
import torch
from torch.nn import Module,Linear
class BatchLinear(Linear):
def forward(self,x):
self.weight.bmm
class SimpleTransformer(Module):
def __init__(self,dim_in,dim_out):
super().__init__()
self.k = Linear(dim_in,dim_out)
self.q = Linear(dim_in,dim_out)
self.v = Linear(dim_in,dim_out)
def forward(self,x):
k,q,v = self.k(x), self.q(x), self.v(x)No, not like this. I need to check out how the batch matrix multiply works.
import torch
from torch.nn import Module,Linear
w = torch.rand(3,2)
x = torch.rand(2,4,2)
r = x @ w.t()
r.shapeActually I am confused now. Why isn't the regular mm good enough?
https://pytorch.org/tutorials/intermediate/seq2seq_translation_tutorial.html
attn_applied = torch.bmm(attn_weights.unsqueeze(0),
encoder_outputs.unsqueeze(0))I am confused. Even with multiple heads it should not be necessary.
11:40am. ...Ah, right. I see. The values should be using bmm. No doubt about it.
11:50am.
def forward(self,x):
q,k,v = self.q(x), self.k(x), self.v(x)
z = torch.einsum('bsa,bsa->bs',q,k)
z = z / torch.sqrt(q.shape[-1])
I am a bit confused about the division by sqrt d. If the purpose is for a linear layer to come up to 1, then it would be better to use the mean.
https://paperswithcode.com/method/scaled
Ah, they are trying to scale the expected variance. I see.
12:15pm.
import torch
from torch import Tensor
from torch.nn import Module,Linear
from torch.nn.init import xavier_normal_
from torch.nn.parameter import Parameter
class SimpleTransformer(Module):
def __init__(self,dim_in,dim_out):
super().__init__()
self.q, self.k, self.q = [Parameter(xavier_normal_(torch.empty(dim_out,dim_in))) in range(3)]
def forward(self,x,mask):
q,k,v = self.q(x), self.k(x), self.v(x)
z : Tensor = torch.einsum('qsa,qsa->qs',q,k) / torch.sqrt(self.k.shape[-1])
z = z.masked_fill(mask,float('-inf'))
z = z.softmax(-1)
torch.einsum('qs,qsa->qa',z,v)Is this right. It feels wrong somehow.
z : Tensor = q.view(dim_batch*dim_seq,dim_el) @ k.view(dim_batch*dim_seq,dim_el).t() / dim_elThis is so complicated.
z : Tensor = torch.einsum('qsa,qsa->qsqs',q,k) / dim_elNo this was not how I imagined it.
def forward(self,x,mask):
q,k,v = self.q(x), self.k(x), self.v(x)
dim_batch, dim_seq, dim_el = k.shape
z : Tensor = torch.einsum('qsa,qsa->qss',q,k) / dim_elYeah, this is where the batch matrix multiply would come into play.
z : Tensor = torch.einsum('qca,qba->qcb',q,k) / dim_elMaybe I could do it like this in order to disambiguate the dimension ordering.
def forward(self,x,mask):
q,k,v = self.q(x), self.k(x), self.v(x)
dim_batch, dim_seq, dim_el = k.shape
z : Tensor = torch.einsum('qca,qba->qcb',q,k) / dim_el
z = z.masked_fill(mask,float('-inf'))
z = z.softmax(-1)
torch.einsum('qs,qsa->qa',z,v)I know how to deal with the mask, but I am confused about the softmax dimension.
12:50pm.
class Attention(Module):
def __init__(self,dim_in,dim_out):
super().__init__()
self.q, self.k, self.q = [Parameter(xavier_normal_(torch.empty(dim_out,dim_in))) in range(3)]
def forward(self,x,mask=None):
q,k,v = self.q(x), self.k(x), self.v(x)
dz, dk, da = k.shape
z : Tensor = torch.einsum('zqa,zka->zqk',q,k) / da
if mask is not None:
mask = torch.logical_or(mask.view(dz,dk,1),mask.view(dz,1,dk))
z = z.masked_fill(mask,float('-inf'))
torch.einsum('zqk,zka->zqa',z.softmax(-1),v)Let me set it like this. Einsum really makes dealing with dimensions a lot easier.
return torch.einsum('zqk,zka->zqa',z.softmax(-1),v)Can't forget the return.
Ah, yes. What will softmax do to an array of -infs?
>>> q.softmax(-1)
tensor([nan, nan, nan, nan])
1pm.
return torch.einsum('zqk,zka->zqa',z.softmax(-1).nan_to_num(),v)So I should fix it using something like this...
1:10pm.
class Attention(Module):
def __init__(self,dim_in,dim_v,dim_qk=None):
super().__init__()
dim_proj = dim_v if dim_qk is None else dim_qk
self.q, self.k, self.v = Linear(dim_in,dim_proj,bias=False), Linear(dim_in,dim_proj,bias=False), Linear(dim_in,dim_v)
def forward(self,x,mask=None):
q,k,v = self.q.forward(x), self.k.forward(x), self.v.forward(x)
dz, dk, da = k.shape
z : Tensor = torch.einsum('zqa,zka->zqk',q,k) / da
if mask is not None:
mask = torch.logical_or(mask.view(dz,dk,1),mask.view(dz,1,dk))
z = z.masked_fill(mask,float('-inf'))
return torch.einsum('zqk,zka->zqa',z.softmax(-1).nan_to_num(),v)Let me set it up like this. This will serve nicely.
Let me take a break here.
z : Tensor = torch.einsum('zqa,zka->zqk',q,k) / sqrt(da)Ah, I forgot the sqrt.
if mask is not None:
z = z.masked_fill(torch.logical_or(mask.view(dz,dk,1),mask.view(dz,1,dk)),float('-inf'))I also remember that assigning to the params will change the defaults. Let me do it like this then."
kthread: add kthread_create_worker*()
Kthread workers are currently created using the classic kthread API, namely kthread_run(). kthread_worker_fn() is passed as the @threadfn parameter.
This patch defines kthread_create_worker() and kthread_create_worker_on_cpu() functions that hide implementation details.
They enforce using kthread_worker_fn() for the main thread. But I doubt that there are any plans to create any alternative. In fact, I think that we do not want any alternative main thread because it would be hard to support consistency with the rest of the kthread worker API.
The naming and function of kthread_create_worker() is inspired by the workqueues API like the rest of the kthread worker API.
The kthread_create_worker_on_cpu() variant is motivated by the original kthread_create_on_cpu(). Note that we need to bind per-CPU kthread workers already when they are created. It makes the life easier. kthread_bind() could not be used later for an already running worker.
This patch does not convert existing kthread workers. The kthread worker API need more improvements first, e.g. a function to destroy the worker.
IMPORTANT:
kthread_create_worker_on_cpu() allows to use any format of the worker name, in compare with kthread_create_on_cpu(). The good thing is that it is more generic. The bad thing is that most users will need to pass the cpu number in two parameters, e.g. kthread_create_worker_on_cpu(cpu, "helper/%d", cpu).
To be honest, the main motivation was to avoid the need for an empty va_list. The only legal way was to create a helper function that would be called with an empty list. Other attempts caused compilation warnings or even errors on different architectures.
There were also other alternatives, for example, using #define or splitting __kthread_create_worker(). The used solution looked like the least ugly.
Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Petr Mladek [email protected] Acked-by: Tejun Heo [email protected] Cc: Oleg Nesterov [email protected] Cc: Ingo Molnar [email protected] Cc: Peter Zijlstra [email protected] Cc: Steven Rostedt [email protected] Cc: "Paul E. McKenney" [email protected] Cc: Josh Triplett [email protected] Cc: Thomas Gleixner [email protected] Cc: Jiri Kosina [email protected] Cc: Borislav Petkov [email protected] Cc: Michal Hocko [email protected] Cc: Vlastimil Babka [email protected] Signed-off-by: Andrew Morton [email protected] Signed-off-by: Linus Torvalds [email protected] Signed-off-by: celtare21 [email protected] Signed-off-by: YousefAlgadri [email protected]
Migrate: Add an identifier to each answerset
When a section can have more than one AnswerSet per Plan, the end-user needs to be able to refer to each AnswerSet. We could use use the primary key, but having a dedicated text-field makes it possible for the end-user to use their own preferred identifiers, like a DOI or a nick name.
TODO: Have a question input type that sets this identifier. This will most likely lead to the information being stored in two locations. However: not storing anything in AnswerSet.data would lead to ugly hacks regarding validity and branching. Not storing it on a direct field on the AnswerSet would lead to other ugly hacks, as in: having to check if the section has a question with the magical input in order to find the question id, then look up the answer. The added complexity would not be worth it.
Merge tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
Pull clk updates from Stephen Boyd: "This round has a diffstat dominated by Qualcomm clk drivers. Honestly though that's just a bunch of data so the diffstat reflects that. Looking beyond that there's just a bunch of updates all around in various clk drivers. Renesas and NXP (for i.MX) are two SoC vendors that have a lot of patches in here.
Overall the driver changes look to be mostly enabling more clks and non-critical fixes that we could hold until the next merge window.
I'm especially excited about the series from Arnd that graduates clkdev to be the only implementation of clk_get() and clk_put(). That's a good step in the right direction to migreate eveerything over to the common clk framework. Now we don't have to worry about clkdev specific details, they're just part of the clk API now.
Core:
- clkdev is now the only option, i.e. clk_get()/clk_put() is implemented in only one place in the kernel instead of in drivers/clk/clkdev.c and in architectures that want their own implementation
New Drivers:
- Texas Instruments' LMK04832 Ultra Low-Noise JESD204B Compliant Clock Jitter Cleaner With Dual Loop PLLs
- Qualcomm MDM9607 GCC
- Qualcomm SC8180X display clks
- Qualcomm SM6125 GCC
- Qualcomm SM8250 CAMCC (camera)
- Renesas RZ/G2L SoC
- Hisilicon hi3559A SoC
Updates:
- Stop using clock-output-names in ST clk drivers (yay!)
- Support secure mode of STM32MP1 SoCs
- Improve clock support for Actions S500 SoC
- duty cycle setting support on qcom clks
- Add TI am33xx spread spectrum clock support
- Use determine_rate() for the Amlogic pll ops instead of round_rate()
- Restrict Amlogic gp0/1 and audio plls range on g12a/sm1
- Improve Amlogic axg-audio controller error on deferral
- Add NNA clocks on Amlogic g12a
- Reduce memory footprint of Rockchip PLL rate tables
- A fix for the newly added Rockchip rk3568 clk driver
- Exported clock for the newly added Rockchip video decoder
- Remove audio ipg clock from i.MX8MP
- Remove deprecated legacy clock binding for i.MX SCU clock driver
- Use common clk-imx8qxp for both i.MX8QXP and i.MX8QM
- Add multiple clocks to clk-imx8qxp driver (enet, hdmi, lcdif, audio, parallel interface)
- Add dedicated clock ops for i.MX paralel interface
- Different fixes for clocks controlled by ATF on i.MX SoCs
- Add A53/A72 frequency scaling support i.MX clk-scu driver
- Add special case for DCSS clock on suspend for i.MX clk-scu driver
- Add parent save/restore on suspend/resume to i.MX clk-scu driver
- Skip runtime PM enablement for CPU clocks in i.MX clk-scu driver
- Remove the sys1_pll/sys2_pll clock gates for i.MX8MQ and their bindings
- Tegra clk driver no longer deasserts resets on clk_enable as it gets in the way of certain power-up sequences
- Fix compile testing for Tegra clk driver
- One patch to fix a divider on the Allwinner v3s Audio PLL
- Add support for CPU core clock boost modes on Renesas R-Car Gen3
- Add ISPCS (Image Signal Processor) clocks on Renesas R-Car V3U
- Switch SH/R-Mobile and R-Car "DIV6" clocks to .determine_rate() and improve support for multiple parents
- Switch Renesas RZ/N1 divider clocks to .determine_rate()
- Add ZA2 (Audio Clock Generator) clock on Renesas R-Car D3
- Convert ar7 to common clk framework
- Convert ralink to common clk framework"
- tag 'clk-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: (161 commits) clk: zynqmp: Handle divider specific read only flag clk: zynqmp: Use firmware specific mux clock flags clk: zynqmp: Use firmware specific divider clock flags clk: zynqmp: Use firmware specific common clock flags clk: lmk04832: Use of match table clk: lmk04832: Depend on SPI clk: stm32mp1: new compatible for secure RCC support dt-bindings: clock: stm32mp1 new compatible for secure rcc dt-bindings: reset: add MCU HOLD BOOT ID for SCMI reset domains on stm32mp15 dt-bindings: reset: add IDs for SCMI reset domains on stm32mp15 dt-bindings: clock: add IDs for SCMI clocks on stm32mp15 reset: stm32mp1: remove stm32mp1 reset clk: hisilicon: Add clock driver for hi3559A SoC dt-bindings: Document the hi3559a clock bindings clk: si5341: Add sysfs properties to allow checking/resetting device faults clk: si5341: Add silabs,iovdd-33 property clk: si5341: Add silabs,xaxb-ext-clk property clk: si5341: Allow different output VDD_SEL values clk: si5341: Update initialization magic clk: si5341: Check for input clock presence and PLL lock on startup ...
More vault treasures (and also a couple of Nudzirath temple ones that got roped in)
Magic-blocking items Antimagic rift: Blocks most types (any types not covered by the other two) of monster spelcasting, blocks your spellcasting if you cast with Int. Catapsi vortex: Blocks monster psionic spellcasting and mind blasts. Blocks your spellcasting if you cast using Cha (which Madmen do) and blocks your psychic powers. Misotheistic pyramids: Blocks monster divine spellcasting. Blocks prayer, sacrificing, altar curse-testing, and priest services -mirrored fragments: Also causes monsters to fracture on death (this is pretty bad). Reading or using it will teach any character Nudzirath's seal. --Includes improvement to Nudzirath; mirrors are shown as options for 't'. -black-stone: Grants dark vision and blocks all other forms of vision. --Includes a modification to Align2gangr such that it will always return the GA_s for your pantheon. Introduces AltarAlign2gangr to handle altars. Probably indroduces some bugs in the process :( -All durations increase exponentially (1.1x) with repeated uses. It is possible to permanently turn off a given type of magic by using enough of a given type of item. The math should be handled to not overflow (eventually stops incrementing the number used, the cutoff is much lower than both the total size of a long int and the max value of a float) --Duration incrementing functions changed to take long durations instead of int durations (these functions then cap what they're given appropriately).
Regen-up items Vital soulstone: Increases health regen while in inventory (stacks), use for burst healing. More effective when cursed. Uses soul coin rules. Spiritual soulstone: Inccreases energy regen while in inventory (stacks), use for burst recovery. More effective when cursed. Uses soul coin rules.
Rod of Force: Treasure for zigguraut chests -2x lightsaber-type weapon using longsword or two-handed sword skill -zap for force bolt + knockback beam (uses 10000 turns of charge/100 hits of charge) -When used unlit, suppresses the special abilities of any monster with spellcasting powers. More effective vs. clerical casters. May cancel clerical casters. Gains 100 charge (1 hit) per turn of suppression damage inflicted, or all charge when cancelling a clerical caster. -When wielded, absorbs force bolts and wand of striking blasts and converts them to charge (10,000 turns of charge/1 zap) --In order to make a weapon that can be zapped, separates oc_dir and dtype into separate fields rather than overloading oc_dir (oc_dtyp introduced. oc_dtyp still overloaded with armor coverage info ): ) --Zap is handled for MvYou, but monster AI doesn't actually do this.
Slightly re-works how hell vault items are created, in prep for placing them explicitly in .des files.
Multiple Texture slots
Fucking Hell, I feel like I'm an idiot? The issue was GameLoop.cpp line 74, where it sets the active texture. Everything I was reading online said you need to do it after loading the texture, but you have to do it before... UGH. Now I get to rewrite my Texture Atlas Class. 3rd times a charm
Add support for doctor examination
Yes folks you read that correctly. There is now a doctor in the house.
No need to be afraid, this is great as it will help identify potential complications that might exist and at least identify various symptons that exist in order to help us understand the health of your integration with our services.
And of course in the spirit of all things we do here, we kept it as a simple, pleasant, painless experience.
More importantly, we have as a result of the doctors examinations, discovered and cured a large number of issues for an overall vastly improved end user experience.
Can I get warm welcome for our newest addition, the tg doctor ^_^
Co-authored-by: Animesh Koratana [email protected]
Add a Code of Conduct to Viceroy
While this commit could really just be the title I think it's important to expand on the "why have a Code of Conduct?" question that gets asked. Sometimes in good faith, but unfortunately often in bad faith. For those who think a Code of Conduct is bad and try to open up issues on the issue tracker about this addition we'll kindly ask you to leave and lock them. For those who are legitimately curious do read on as to the why.
Often times tech is a pretty hostile space for many people, especially with Open Source Software (OSS), but also at companies themselves. Now you might be thinking "Well I've never experienced harrassment or rude people in tech" and that's great. Honestly I'd hope no one would be subject to that kind of treatment. That being said it's annecdotal and if you spend time talking to minorities in tech (whether it's sex, race, gender identity or something else) you'll find this is often not the case. Personally I've barely experienced anything at the level of harrasment that coworkers or friends in the industry have experienced. We want a Code of Conduct so that we have a mechanism to enforce safety within our community. While we'd like to be open for everyone we don't want to be open at the expense of the well being of others. Working on code just isn't fun if all you're going to recieve is angry threats from developers and worst of all if we tolerate that kind of behavior people will leave and they won't make a fuss of it they'll just go.
Worse, people see this behavior being tolerated and won't want to be part of that community or participate in it. By tolerating intolerance by others, a community either diminishes or becomes intolerant itself. This is where the paradox of intolerance comes in, in order to have a tolerant and welcoming community we must be intolerant of intolerance. This is what the Code of Conduct is for! We don't want people who are rude or make people feel unwelcome. We want others who are kind, caring, and make others want to be here and feel supported. A Code of Conduct is by no means perfect and only one part of this, but it's the bare minimum needed to have a healthy community.
If you feel excluded because of the Code of Conduct, I'd ask you to think why. Is it because you can't be rude towards someone you probably don't know online? Is it because in a public professional space you can't swear? Is it so bad that you can't do these things? The Code of Conduct doesn't ask for a lot and while you might say "We're all adults here, we don't need one", consistently that just hasn't been the case in this industry. We're always open to contributions we just want to have a certain level of professionalism.
At Fastly we also have our own values and code of conduct and part of this includes the values "We are trustworthy" and "We are good people" and if we want others to trust us and if we want to be good people looking out for our community's well being over the hurt feelings of an individual who feels excluded because they can't be rude is a tradeoff we'll make in a heartbeat.
In summary while a Code of Conduct does exclude people, it excludes the kinds of people we'd rather not interact with and instead invites the people who we would want to interact with into the community. We want people who care and who are kind and compassionate. If you've read this long do come join us. A better world of software where we are nice to each other, where we build each other up, and build some really neat things is possible.
Further fix ALTER COLUMN TYPE's handling of indexes and index constraints.
This patch reverts all the code changes of commit e76de8861, which turns out to have been seriously misguided. We can't wait till later to compute the definition string for an index; we must capture that before applying the data type change for any column it depends on, else ruleutils.c will deliverr wrong/misleading results. (This fine point was documented nowhere, of course.)
I'd also managed to forget that ATExecAlterColumnType executes once per ALTER COLUMN TYPE clause, not once per statement; which resulted in the code being basically completely broken for any case in which multiple ALTER COLUMN TYPE clauses are applied to a table having non-constraint indexes that must be rebuilt. Through very bad luck, none of the existing test cases nor the ones added by e76de8861 caught that, but of course it was soon found in the field.
The previous patch also had an implicit assumption that if a constraint's index had a dependency on a table column, so would the constraint --- but that isn't actually true, so it didn't fix such cases.
Instead of trying to delete unneeded index dependencies later, do the is-there-a-constraint lookup immediately on seeing an index dependency, and switch to remembering the constraint if so. In the unusual case of multiple column dependencies for a constraint index, this will result in duplicate constraint lookups, but that's not that horrible compared to all the other work that happens here. Besides, such cases did not work at all before, so it's hard to argue that they're performance-critical for anyone.
Per bug #15865 from Keith Fiske. As before, back-patch to all supported branches.
Discussion: https://postgr.es/m/[email protected]
Gehennom treasure upgrades
Valley of the Dead: one of each coin type, randomly placed.
Juiblex (demon lord): 6x wage of gluttony, 12x demon vault loot.
Zuggtmoy (demon lady): 6x wage of gluttony, 12 demon vault loot.
Yeenoghu (demon lord): 3x wage of gluttony, 3x wage of wrath, 12x demon vault loot.
Baphomet (demon lord): 3x wage of pride, 3x wage of wrath, 12x demon vault loot
Kostchtchie (demon lord): 3x wage of wrath, 3x wage of sloth, 12x demon vault loot
Malcanthet (demon prince): 3x wage of lust, 3x wage of envy, 18x demon vault loot
Graz'zt (demon prince): 3x wage of pride, 3x wage of lust, 18x demon vault loot
Avatar of Lolth (demon prince): 2x wage of pride, 2x wage of lust, 2x wage of envy, 18x lolth vault loot
Pale Night (demon prince): up to 36x tannin vault loot at various probabilities -Pale Night buffed with a summon tanninim spell (1/36 casts, summons shalosh or teraphim. -Pale Night gains acid resistance (so her own shalosh can't kill her)
Orcus (demon prince): 3x wage of pride, 3x wage of sloth, 18x demon vault loot.
Dagon (tannin prince): up to 36x tannin vault loot at various probabilities scattered in the water. -Dagon buffed with khaamnun pull and drain level attacks
Demogorgon (prince of demons): 3x wage of pride, 3x wage of envy, 6x ancient vault loot, 6x angel vault loot, 6x tannin vault loot, 6x devil vault loot, 12x demon vault loot.
Lamashtu (prince of demons): 3x wage of envy, 3x wage of greed, 12x angel vault loot, 12x tannin vault loot, 12x demon vault loot. -Lamashtu armies supplemented with angels.
Bael (archdevil): 6x wage of wrath, 3x wage of envy, 9x angel vault loot, 9x devil vault loot, 9x demon vault loot.
Dispater (archdevil): 3x wage of greed, 3x wage of envy, 3x wage of pride, 6x ancient vault loot, 3x angel vault loot, 18x devil vault loot.
Mammon (archdevil): a whole mess of wages of greed, 27x devil vault loot.
Belial/Fierna/Naome (archdevils): 3x wage of lust, 3x wage of pride, 3x wage of wrath, 27x devil vault loot, 9x75% devil vault loot.
Leviathan has no loot at present. Loot should be included when "opening" the creature in the ice.
Lilith (archdevils and court): 6x wage of envy, 6x wage of lust, 6x wage of pride, 27x devil vault loot.
Baalzebub (archdevil): 6x wage of pride, 3x wage of sloth, 18x angel vault loot, 9x devil vault loot.
Mephistopheles (archdevil): 3x wage of pride, 2x wage of greed, 2x wage of envy, 2x wage of wrath, 27x devil vault loot, 9x75% devil vault loot. Baalphegor (ancient of hell): 4x antimagic rift, 4x catapsi vortex, 24x ancient vault loot, 8x75% ancient vault loot.
Asmodeus (archdevil): 3x antimagic rift, 3x catapsi vortex, 3x wage of pride, 27x devil vault loot, 27x90% devil vault loot, 27x45% devil vault loot, random soul coins with the treasury golems.
README.md
My name is Juan, I'm 28 years old, and I live in La Plata, Buenos Aires (ARG).
- 🎓I have a degree in Bussiness, and I am studying IT. Yes, I love learning!
- 💡 I complemented my background with Data Analytics and Data Science studies.
- 📈 Currently working as an analyst of planing and control, at EDELAP (energy industry).
- 📫 If you want to contact me: [email protected]