Recently I saw these test results on huggingface.co:
Now, the large language models (LLMs) are nearmost reaching the benchmark of average humans.
On huggingface.co/../open_llm_leaderboard large language models are tested against certain well-defined key benchmarks using an unified framework by applying a set of distinct evaluation tasks.
The details each part of the test are shown below:
The top performer, i.e. those having the top improvements are the maths related tests:
GSM8k (5-shot) – diverse grade school math word problems to measure a model’s ability to solve multi-step mathematical reasoning problems.
MMLU (5-shot) – a test to measure a text model’s multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more.
Let’s see which parts of the test have reached the highest score:
Winogrande (5-shot) – an adversarial and difficult Winograd benchmark at scale, for commonsense reasoning.
The common sense reasoning test is now at the leading edge, showing up 90 of max. 94 points! Congratulations!
Modern physics made a fool of itself, did follow Einstein blindly – as Max Planck directed. And now? They are none the wiser, so klug als wie zuvor. There was given a theory that pretended to get the c-is-const-hypothesis compatible with relativity principle. But it failed from beginning and the interpreters tried to make the students understand the paradoxes – until today. By applying high-tech experiments around the Einstein believing globe they try to prove that -despite all paradoxes- the theory of Relativity does hold.
What humans cannot understand they cannot take as useful. What is a theory good for if humans do not understand it?
After adjustment the instruments measure what the theory predicts: Modern Michelson-Morley-Experiments seem to prove that there is no ether and light speed is constant at all diections – independent from motion.
General relativity and quantum theory are experimentally justified theories describing nature. Despite enormous efforts, a commonly accepted theory was not yet found.
If the so-called Lorentz Invariance holds the speed of light must be measured to be the same in any direction, and during any contínuous constant motion as well.
In our previous experiments a possible anisotropy of c could be restricted under a level of 5 · 10-16. Our current setup shows an improvement of more than one order of magnitude and provides SME-parameters at the level of 10-17.
Special relativity (SR) underlies all accepted theories of nature at the fundamental level.
In the classic setup, one compares the speed of light c in two orthogonal interferometer arms by observing the interference fringes. If c depends on the direction of propagation, the fringes would move if the setup is rotated. A modern technique measures resonance frequencies of two optical cavities. Violations of the isotropy of c may be detected as variation of the frequencies due to rotations.
A significant consequence of Lorentz symmetry is the isotropic nature of the speed of light, which remains invariant under rotation. Measuring the isotropy of the speed of light has played an important role in physics, starting with the seminal Michelson and Morley interferometer experiment in the late nineteenth century.
Question:
If we don’t want to measure some impact of rotational speed, how fast should the table rotate?
It should rotate as fast as possible!
Fallacy 1: Don’t ask the reason why
SRT just stated that speed of light is constant and therefore c must be measured to be const. But SRT did not explain how the light fronts would manage this constness – despite the fact that the light paths are different in length. The Lorentz contraction did not help H.A. Lorentz to explain, not A. Einstein to understand, not W. Pauli and not Richard Feynman – the latter just said: Must believe it.
Fallacy 2: Faith instead of critics
Is there one professorial chair in theoretical physics held by a person that is especially sceptically about Einstein’s theories?
Fallacy 3: Working without craftmenship
Craftmen need to prove what they are implementing. For this to realize they need to apply their own basics of understanding as their very invariants. The researcher at Humboldt University (Optical Metrology) just use the constness hypothesis of Albert Einstein to calibrate their instruments – without knowing the parameters that might cause or hinder that constness.
Euclidian Geometry to explain Lorentz Invariance
At first, we need to explain:
If light goes independently (from sender, receiver etc.) while the receiving mirror (beam-splitter etc.) is in motion there must be an adjustment of the reflector to match both beams.
The general assumption on this -always ignored- detail is that light goes the Fermat’s path which is equal to the shortest distance. If we use the arm of LIGO interferometer of 4 km length this detail no longer may be ignored: While the light passes 4000 meters the earth rotates forward by ca. 30cm; the beam would miss the target. Therefore the mirrors of the cavity need to have a concave curvature.
LIGO’s pure fused silica mirrors – 40 kg each. (Credit: Caltech/MIT/LIGO Lab)
If spherical mirrors are used the light paths will change.
The path of light takes an asymmetrical form when spherical mirrors on our moving earth are used: The mirror (indicated by the circle) moves during the light passes (from A) upward to B matching it at the red dotted position. The reflection point is offset, the distance to the target is shortened a bit and consequently, the reflection becomes asymmetrically.
According to Fermat’s law the shortest path will be taken by light. Therefore the asymmetric light path is the one to be used in reflection-in-motion case!
Calculating the distance between both arms of the Michelson-Morley-Interferometer shows that both arms have exactly equal length if both beams are focused at the matching position that shows stable interference patterns.
Summary:
In order to prevent the walkoff of the vertical beam you have to use a spherical mirror.
In order to get the light beams joined after splitting you must adjust the mirrors properly.
In order to get a stable interference pattern you must adjust the mirrors during rotation.
If the light shows stable interference during rotation both mirrors will be set to equal path lengths and the vertical beam uses the asymmetric light path.
The operator himself selects the best matching pair of light paths. And – this is new for the scientific people selling Einstein’s relics – there is a pair of equal length.
If near-most equality is not enough for your setup, just use triple-reflectors instead!
It was John Smith, who did the delivery of the monkey wrenches, according to the author of Database Design and Relational Theory, Chris Date. He wanted to show the formal anomaly (connection trap) that is accompanied by loss of information if n-ary relations are decomposed into multiple binary relations and rejoined again. What he omitted was to indicate why there is a need to use a n-ary relation instead of binaries – without any formal reason. The formal oriented database design does not give a clue on design, i.e. what form to use for what. This all has to be provided by domain knowledge, i.e. the semantics and pragmatics that provide the detailed structure of the objects to depict.
The SPJ-relation “Supplier-Parts-Project” contains the data of all actual deliveries of parts for any project. By decomposing SPJ into three binary relations (Supplier-Parts, Parts-Project and Supplier-Project) the relations are projected such that edges (pairs of nodes) only become available. Joining relations means to combine all tuples by identity. The join operation creates the set of all feasible tuple combinations, not just the subset of the real ones. The example above shows four deliveries instead of three.
The idea to use an uniform structure for all data has been proposed by J.R. Abrial (“The Semantic Binary Model, J.R. Abrial, “Data Semantics,” in J.W. Klimbie, K.L. Koffeman (eds.), Data Base Management. North Holland, 1974.”) Within Abrial’s framework the SPJ-relation would be redesigned as three binary relations with foreign keys to some new relation containing the keys of all delivery events. One could easily reconstruct the facts by joining along the delivery keys. From this point of view binary restrictions may effect the efficiency of the retrieval process but cannot restrict the information to be retrieved.
After aggregation there may be anomalies…
We know the relativistic point of view which is all around during high school, at university and nowadays in the public press as LIGO’s search for gravitational waves is featured. The relativistic point of view does not accept a predefined common frame of reference and insists to apply relative velocities only. This is a mystery! How did Einstein manage to add the velocities, e.g. the speed of A and the speed of C? If vAB is said to be the speed of A with respect to B then vBA is the speed of B with respect to A. If vBC is to be added to vAB the result would more than doubtful because vBC needs C to refer to some anchor and vAB refers to B which is moving independently from C. So we must state:
The addition of velocities without common reference is void.
To add vBC and vCB is void then. To add vBC and vAB we have to use B as common reference. But what is the semantics of this sum if we don’t know the angle ABC? We have to admit:
The addition of velocities without regarding the incident angle at the common reference is void.
How do we get the angle ABC, and how do we ensure that it does not change during the maneuver? There must be a measuring process in order to check the distance between A and C -or which does the same- to measure the speed vAC directly. So we have:
In order to add two relative velocities we must havemore than two relative velocities, we must have all relative velocities.
If all points are lined up and the point in the middle hides the third from the first the angle could be checked easily. But all other cases show the same rule: If we want to aggregate the velocities of all particles within a given segment we have to know the complete graph which is the velocity of any node with respect to any other node. This is equivalent to be omniscient.
What did science wrong when dealing with Einstein and his followers? They never tackled the fundamental lack of inference which opens the gateway for paradoxical phenomenons that may not even concerned as antinomies.
The relativistic position implies to be omniscient.
gg:sfit 