The main reason for the Soviet losses late in the war is that attacking is many times more costly than defending. Against a 'comparable' enemy (im using the term loosely here), attacking a fortified position requires over 3:1 ratio in tanks, aircraft, artillery, men, etc. simply because a vast amount of that is going to be lost. Defensive power, via machine-guns and AT-guns, is increased by several factors, as compared to offensive power.

This is compounded by the standard Soviet doctrine, the deep operation, which stressed massive combat operations engaging the enemy throughout his while operational depth simultaneously, with supporting pinning attacks, in order to destroy the opponent, not simply rout them. Destroying (encircling most often) the enemy is alot more difficult than simply causing a rout, and thus requires a higher investment of men and material.

The key part of the deep operation, and the reason for the high casualties on the Soviet side, is that if you engage the enemy throughout his entire depth, the forces you have engaged also shoot back at you, (if the enemy is in range, so are you) and couple this with the high reliance on the MG-42 on the German side (if I remember correctly, it was 3 42s pr platoon) and as was shown in WW1, charging into machine-guns is a costly affair, only partially offset by the usage of tanks which was for the most part relegated to be used in deep exploitation groups. (Usually a tank corps or a cavalry-mechanized group)

As for why the losses were lower on the western front, the doctrines were different, the fighting was different, and most importantly, the numbers were different. (Operation Overlord launched with 1.5 million troops vs 350.000, Bagration had 2.5 million vs 850.000, give or take)

If one takes a look at the relative advantages of the combatants on the allied side, the Soviets simply had more men to throw into the grinder, and were markedly more willing to throw said men into said grinder. Although the US had a vast manpower reserve, they were unable (or unwilling, im not an expert on the western front) to utilize said manpower, instead choosing to leverage the enormous industrial might of the arsenal of democracy, choosing quality over quantity (simplification).

To sum up:

1. Attacking is always more costly than defending. (You could hardly call the circumstances of 1941 'defending').
2. The Deep Operation which stressed massive simultaneous attacks incurred a large cost in lives for the success that it achieved. It was tailored to the unique Soviet situation as well as the Soviet geography which would have made pushing back the Germans conventionally take a looooong time.
3. The Western Front was a different beast to the Eastern Front. Lower intensity of fighting, lower numbers, higher amount of high-tech material.

Sources:

When Titans Clashed: How the Red Army Stopped Hitler - Glantz & House
Fundamentals of the Deep Operation - Isserson
Second World War - Beevor
Russian Way of War - Harrison

As the number of upvotes suggests, this is a very interesting question and one without a cut and dry answer. The simple, almost cop-out answer is that for 1944 and 1945 the Soviets were on the offensive, and casualty ratios often skew in the defender's favor. However, this doesn't really do the topic justice. Before diving in, I will throw the caveat that TiK is not a historian and should probably be trusted about as far as you can throw him. He pulls from good sources, most of the time, but I've noticed that his videos are very much "I am regurgitating whatever the most recent book I read said" and he's a bit of a nutter anyhow.

So, how do we explain the disparity in casualties during the third period of the war? Some of it is indeed reporting disparities: the Soviets and the Wehrmacht seem to have accounted for wounded in action differently, and German wounded-in-action numbers will vary somewhat depending on whether you're going from Wehrmacht or Heeresarzt (military medical service) records. The Heeresarzt numbers are known to lag somewhat, so it is possible (although difficult to confirm) that there is a 'tail' of a couple of weeks between the end of a battle and when all the casualties from a battle are tabulated.^(1) This would lead to some funny casualty ratios depending on how the Wikipedia editors define the end of the battle. It is notable that the linked articles are generally comparing the raw Wehrmacht and/or Heeresarzt data to the studies done by Krivosheev on Red Army casualties. There may be a sort of apples and oranges comparison happening here, where Krivosheev is tabulating perhaps 60 days worth of casualties while the wiki editors are tabulating 50 days worth, depending on definitions of the start and end of the battle. This is something I've run into recently with casualty figures at Stalingrad, for instance. I don't have a copy of Krivosheev's Soviet Casualties and Combat Losses in the Twentieth Century at hand to confirm this definitions question, unfortunately.

There is also the matter of figuring out how each side counted their casualties - all wounded counts are not created equal. For starters, I am having trouble determining thus far how complete the Heeresarzt numbers are. Namely, soldiers who were lightly wounded and were treated at their position or in a Verwundetennest (a casualty collection point in American parlance) by a field medic, but did not require treatment at the Truppenverbandplatz (battalion aid station) - would they be reported in any official capacity? My guess is that in the late stages of the war in particular, as the availability of trained medical personnel as well as supplies and casualty transport infrastructure worsened, only more severe injuries would be prioritized for treatment at rear facilities. If a casualty never made it to the Truppenverbandplatz, then it would likely be up to his chain of command as to whether he was reported or not; this introduces inconsistencies. However, I must emphasize that this is a conclusion I arrive at myself based largely on the description of Wehrmacht medical care in the US Army Office of Medical History's official report, and I really hope one of this sub's Wehrmacht experts can point me in the direction of a more comprehensive source. What I can say with somewhat more certainty is that, by 1944-45, the Soviet field medical system was much more mature than it was at the beginning of the war. By 1943, there were designated medical squads at company level, responsible for evacuation of casualties to higher care echelons - not too different from the Wehrmacht.^(2) However, while the Wehrmacht's logistical situation was getting worse during the late war, the Red Army's situation was improving, and there was within the GVSU (the Red Army medical branch) a real drive to keep accurate statistics & meet official goals. Efforts were made to involve the GVSU in planning major operations, and from late 1943 onward there were more orderlies & auxiliary orderlies allocated in each rifle division, which would increase the ability of the division to transport casualties to aid stations.^(3) From all this - especially the Soviet military's predilection for getting and using as much statistical data as possible in the planning process - it is reasonable to infer that the Red Army was probably reporting more accurate numbers for wounded-in-action than the Wehrmacht by late 1944.

So, alright. We've hemmed and hawed a bit about whether the Wikipedia casualty ratios are a good comparison, but at the end of the day it is quite clear that no matter what the exact numbers are, the Soviets did often suffer worse casualties in 1944-45 than the Wehrmacht did. What might this tell us about the relative 'combat effectiveness' of the two armies? Fortunately, someone has already done the math for me on this one. These two tables show us the relative combat effectiveness of the Soviet attackers and the Wehrmacht defenders.^(4) This chart shows us some really interesting information. First and foremost, although the raw-number casualty ratios look very lopsided in favor of the Germans, the casualty ratios as a portion of each side's manpower are actually not very favorable for the Germans at all! In one battle within the larger Dnepr-Carpathians Operation which you cite, despite the fact that the Soviets took more casualties than the Germans in raw numbers, as proportions of the overall force the Soviets take 34% casualties (all types - KIA, WIA, MIA) while the Wehrmacht takes 55% overall casualties. How can this be? Well, as much as the trope of Russian numerical superiority is much overplayed, the simple fact is that by 1944 the Soviets are making their major strategic offensives with more men concentrated on that one operation than the Germans have on the defense. I will address this in more depth later, but this is a cornerstone of Soviet doctrine. For the German defenders, this is a comparatively target-rich environment. In other words, even though the proportional casualty ratios are often in favor of the Soviets here, the fact that there are more Soviet soldiers on the battlefield means that more of them will become casualties in total, especially when looking at the 'efficiency' advantage for the defender.

END PART ONE

Good question. There are several points to be made here.

1. Casualties for Bagration are misleading

At first Glantz glance, the Soviets seem to have it worse at Operation Bagration, but upon closer inspection, the majority of Soviet casualties is injured, and the majority of German casualties are either killed or captured/missing. Now injured personnel can be permanently disabled, but a fair portion will either return to active duty or be re-integrated behind the frontline or in society. Captured soldiers and missing ones (note that as the frontline moved, these missing soldiers would find themselves hundreds of kilometers behind enemy lines) on the other hand, are pretty much gone. Many captives would either die in death marches, or be stuck in a POW camp, and there's only occasional stories of people escaping POW camps and walking home, because they're exceptional. This is the big advantage of a successful offensive and something that the Soviets had been on the receiving end of quite extensively during Barbarossa.

2. Lack of overwhelming force on the Soviet side

At Dnieper-Carpathian -at an even later time- the disproportionate casualties can be largely attributed to the lack of overwhelming force on the Soviet side. Rather than the typical 2:1 or even 3:1 local advantage that the Soviets usually secretly built up in preparation for an offensive, Dnieper-Carpathian had 'only' had a 10% advantage in pure numbers.

These two soften the numbers a little bit, but would be no more than a bad excuse, if it weren't for the most important point:

3. The Soviets were on the offensive, against a (still) tough opponent

This is arguably the most important reason. At no point in 1944 and 1945 were the Axis forces not strategically on the defensive on the Eastern Front. Ergo the Soviets were continuously on the offensive. Early in the war they were also desperately trying to be on the offensive, as their doctrine dictated they should, but wholly unable to. By 1944 they had gotten immeasurably better at it, but being on the offensive was still a very costly affair in this time period. In more or less equal battles, breakthroughs needed overwhelming force to guarantee operational success and typically incurred far greater average casualties than defensive actions. The only times this was not the case was in spectacularly successful offensives or against much weaker opponents. The Battle for France and Barbarossa and such were in a way just shocking exceptions, that showcased how highly functional the German army was and/or how dysfunctional the Allied/Soviet armies were, rather than prove that the fundamentals of warfare had changed to make the offensive cheaper than defense in terms of lives.

In this regard, it was fully to be expected that the Soviets, now on the offensive, would pay a heavy price for any successes or reconquest. And we should not underestimate what they were up against at this point either. The common narrative is that the Axis front at this point had been worn out, that the forces were incompetent Italians and demoralized Romanians, and that the German forces had lost vast amounts of trained manpower they absolutely couldn't miss, were increasingly outmatched in quantities of materiel, gradually lost control over the air, and devolved into an outdated doctrine of static defense. And this isn't completely wrong. But it remained a coherent and strong fighting force, up to the last months of the war. The German army, when it was full-strength, had obliterated equal forces. Being no longer full-strength it still proved a force to be reckoned with. You rightfully draw a comparison with the Western front, but consider that the Allies on the Western front had had -to put it bluntly- all the time in the world to build up strength, to prepare, and to train their troops. And yet, the rather small portion of the battered German army that was sent West, held them off for a long time. In addition, contrary to the Western Allies, the Soviets were almost just as worn out as the German forces. Contrary to the popular view, they were very much strained for manpower, and a large portion of their frontline soldiers were green (courtesy also of the immense casualties each unit took during the continuous offensives).

So in conclusion, the casualty ratios of late war operations may appear fairly bad, but they could arguably could have been a lot worse given the state of the two armies, and the offensive nature. The difference in deaths compared with the Western Front is easily explained: The Western Allies fought battles that were either concentrated and minuscule in comparison (Invasion of Italy), or very late in the war (in Normandy and onward), when the German forces were far more battered, almost always bringing with them an overwhelming, extensively trained and lavishly equipped force to fight a smaller force.

With regard to the Western Front, in Normandy the resistance was every bit as stiff and so were the casualties it created. The losses of forces in the beach-head and the bocage were up to par with the very worst WWI figures. After Normandy Germany hemorrhaged manpower it never truly replaced, but where it could it fought with ferocity, if not skill, and did produce as much of casualties as it was in it to produce. In 1945 after the encirclement in the Ruhr there were very few forces left outside the Berlin lines, so where there is no force to offer resistance and no civilian suicide squads, obviously there will only be casualties from car accidents if that.

In the East, the forces involved were much bigger, and the Germans knew extremely well what that murder spree in 1941-2 was going to get them when Stalin's legions stormed their way west. They were quite literally afraid that the kind of sustained mass terrorism they unleashed was going to be 'if Cain sevenfold, then Lamech seventy-sevenfold' visited upon them. To a degree it was.

Added to this, the Soviets were on the offensive in 1943-5, and took the according losses to face vast numbers of German forces that fought and died as a testament to ferocity and the indoctrinating power of a totalitarian regime, if not to tactical, operational, or strategic logic. The Siege of Budapest and the Courland Pocket, two of the bloodiest battles of the late war, were also utterly senseless exercises in the Germans sustaining a campaign long past the point where a rational regime would have cut its losses and moved manpower elsewhere that would have actually made a difference.

As far as why the German casualties were lighter, it's worth noting that the forces in Bagration and the Jhassy-Kishinev Offensive tended to encircle German forces and remove them completely from the order of battle. Too, there's an often-neglected element that 'casualty' and 'fatality' tend to be separate things. The Soviets might have sustained massive casualties but the number of soldiers killed within that category tends to be rather smaller. As it would also be with the Germans, to be frank.

Then, on top of the rest, in 1945 the Soviets invaded densely urbanized territory fought over again with ferocity if not sense. Urban battles take corps and produce oversized battalions as a general rule. Urban battles fought as a deliberate exercise in attrition do so still moreso. Berlin and Budapest were the grandest examples of this pattern, though again Germany just got from that the satisfaction of bumping up the death toll for both sides but nothing else to show for it. The Budapest fighting literally devoured the rest of that 1944 production the Wehraboos make a big fuss about, in all the senseless attempts to relieve the siege the Soviets chewed up and spat out without it having any notable effect on the siege, either stopping it or slowing it down.

Germany could not afford to keep losing entire armies into the Gulag and to create new armies as large as the ones that the USSR was destroying on the battlefield. Germany especially could not afford the loss in combat experience and the ability of its forces to do more than die in place, or even their dying in place to begin with. Of course the casualties of 1941-2 took their own blood toll from the USSR and showed through the war in the relative tactical proficiency of any given Soviet formation outside the Guards forces, but at the end of the day the USSR was able to afford its losses and Germany could not afford its.

As far as the northern sector of the front is concerned, the Soviet forces there never developed the proficiency at major combat operations the forces in the center and the south did. This is at least partially the difficulties first of the Siege of Leningrad, which maximized small group combat over large group combat, and at least partially that the Soviets did not send their best generals to the north where there were fewer decisive advantages to be gained there relative to the center or the south.

This is illustrated in a gruesome fashion in the Courland battles, as well as in the Novgorod Offensive. The adjustment from bloody small-unit battles, an area the USSR never reached parity with the Germans or democratic armies in, to larger-unit battles overtaxed the abilities of the generals commanding the northern and northwestern Fronts. Add to it that the terrain of Northern Russia and the Baltic is not favorable to Soviet strengths and that question is straightforward enough.

Too, cases like the Siege of Breslau, which surrendered on VE Day, illustrate that ferocity on the Soviet side and the willingness to sustain losses that no democratic army could have sustained have limits as far as abilities to get things done. That this resistance did not win the war for the Germans likewise shows that Patton was right when he said nobody wins a war by dying for his country, and that it requires making the other son of a bitch die for his country.

In any appraisal of late-war casualties, the senseless fanaticism in cases like Budapest and Berlin, which account for a preponderance of the death toll, has to be factored in. No democratic army could have gained that willingness to die for no good reason.