Upload Chimera 510M at step 162000

README.md

@@ -33,7 +33,7 @@ language:
 ![Training Curves](training_curves.png)
 
 ## Training
-- **Step:** 140,000
+- **Step:** 162,000
 - **Data:** Mixed (75% FineWeb-Edu, 18% StarCoder, 5% FineMath, 2% UltraChat)
 - **Optimizer:** Muon + AdamW (decoupled embedding LR)
 - **Schedule:** WSD (Warmup-Stable-Decay)

config.json

@@ -26,7 +26,7 @@
   "architecture": "Chimera",
   "config_class": "ChimeraConfig",
   "topology": "4 bottom + 4x3 top = 16 virtual",
-  "step": 140000,
+  "step": 162000,
   "total_params": 510217280,
   "size_label": "510M",
   "model_type": "zara-ml"

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0a152ebb12c8d8b1286d3e4a69a8d4627e3f48b962adf72cb08d35b526729f1a
+oid sha256:f360d5cbdd67918432f8f75650644154322fef74533c6870579162cff1921923
 size 1487197656

training.log

@@ -2246,3 +2246,329 @@ step 142000/250000 | loss 2.8230 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,12
   >>> [Scientists have discovered] Scientists have discovered a protein that regulates the activity of the enzyme that enables the enzyme to break down protein during the digestion process. The protein is shown to regulate a unique metabolic pathway that drives the breakdown of amino acids, which are critical building blocks of proteins.
 A team of scientists from the University of Florida and the University of Florida have discovered that this protein plays a critical role in the formation of proteins, known as peptides
 step 142200/250000 | loss 2.8328 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,122 tok/s | epoch 1
+step 142400/250000 | loss 2.8307 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,125 tok/s | epoch 1
+  >>> val_loss: 3.1400 | bpt: 4.5301 | true_bpb: 1.4573
+  >>> [The] The rudimentary method of administering a drug (usually a medicine) is called hyphensitization. A hyphensitizer is a medicine that helps to decrease your body's response to an infection. Hyphensitizers help to reduce the level of toxins in your body.
+If you need hyphensitizer treatment, it is best to consult your doctor or pharmacist before trying it.
+
+  >>> [Scientists have discovered] Scientists have discovered a previously undiscovered mechanism that could be the key to new therapies to fight cancer and Alzheimer’s disease. Researchers at the University of St. Charles have discovered a new way to attack the disease-causing form of the cancer cell, a way that has traditionally been used to target key genes (genes that control cell growth).
+This work could be a key to slowing the spread of many types of cancer,
+step 142600/250000 | loss 2.8297 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,123 tok/s | epoch 1
+step 142800/250000 | loss 2.8358 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,126 tok/s | epoch 1
+step 143000/250000 | loss 2.8359 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,130 tok/s | epoch 1
+  >>> val_loss: 3.1389 | bpt: 4.5284 | true_bpb: 1.4567 *BEST*
+  >>> [The] The persăn, meaning "courted" in Hebrew, means "tree" in Hebrew. This tree was then "tossed away" by the Israelites. The Jews were not very sophisticated when they were in the market place because they had a great deal of olives.
+24. The Bible says, "From the tree of the Lord" it means "to the right" or "to
+  >>> [Scientists have discovered] Scientists have discovered new genes that regulate the immune system and may potentially help strokes and heart disease.
+"Our study suggests that a variety of receptors might be involved in these diseases, and could help to sweep away some of the underlying causes of these diseases," says the study's senior author, Hiroshi Choi, the researcher's senior author and a professor of biology and genetics at the University of British Columbia.
+The study included more
+step 143200/250000 | loss 2.8345 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,123 tok/s | epoch 1
+step 143400/250000 | loss 2.8245 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,127 tok/s | epoch 1
+  >>> val_loss: 3.1382 | bpt: 4.5275 | true_bpb: 1.4564 *BEST*
+  >>> [The] The Bryonius plant is also classified as Acer pungens.
+For a detailed list of the species and species names of the plant, see the Coleonius Plants page.
+Douglas's flower is the flowers of the Australian Botanical Society.
+When a flower is white, it is called "full-grown".
+Greenish flowers are called "giant flowers".
+The leaves of the Australian Botanical
+  >>> [Scientists have discovered] Scientists have discovered that the most common method in which scientists can create DNA from human cells is through DNA reprogramming. The technique involves using DNA from an individual’s DNA.
+In order for DNA to be “reprogrammed,” scientists must first have the correct DNA sequence. Once this circuit is established, the DNA is then passed through the human DNA coding system. This system allows DNA to be truly cloned into a particular
+step 143600/250000 | loss 2.8331 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,121 tok/s | epoch 1
+step 143800/250000 | loss 2.8220 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,124 tok/s | epoch 1
+step 144000/250000 | loss 2.8171 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,128 tok/s | epoch 1
+  >>> val_loss: 3.1378 | bpt: 4.5269 | true_bpb: 1.4563 *BEST*
+  >>> [The] The oldest and most famous American writer is John Stuart Mill. In a letter to his admirer John Stuart Mill, he wrote that he loved older than frequently.
+This is a good quote. “Why do I still love you?” it is an old question. Mill, considered one of the most intelligent and serious writers of our time, is no more than a rubber first.
+If you are simply looking at
+  >>> [Scientists have discovered] Scientists have discovered a group of enzymes that are also found in soft tissue called neural tissue, and believe that they may be a major contributor to developing dementia, says Dr. Yates.
+“By studying the neurons of patients who have started to develop their problems, we can better understand the biology of the pathology that drives them,” he says.
+Research from the UCSF Center for Brain Research has found that a “fast
+step 144200/250000 | loss 2.8240 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,122 tok/s | epoch 1
+step 144400/250000 | loss 2.8373 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,125 tok/s | epoch 1
+  >>> val_loss: 3.1369 | bpt: 4.5255 | true_bpb: 1.4558 *BEST*
+  >>> [The] The Washington Toileve Court is synonymous with the use of the term “youth.” The City School is the only public school in the District of Columbia that has an inclusive character. Students in the school are the “rural” races, the “feminine” races, and the “shining flag.” The bicentennial celebration of Washington, D.C. is typically held after Parks,
+  >>> [Scientists have discovered] Scientists have discovered that when Australian waterways are saturated with the nitrogen nitrogen lies a significant increase of acid gases which are more poisonous to fish. Polluting waterways with nitrogen nitrogen can cause a disruption of the natural balance of oxygen entirely, and can harm the health of fish and other marine fauna.
+Also, air pollution can affect the oxygen levels in the waterways which means more oxygen is being added to the water.
+step 144600/250000 | loss 2.8199 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,119 tok/s | epoch 1
+step 144800/250000 | loss 2.8175 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,122 tok/s | epoch 1
+step 145000/250000 | loss 2.8187 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,126 tok/s | epoch 1
+  >>> val_loss: 3.1362 | bpt: 4.5245 | true_bpb: 1.4555 *BEST*
+  >>> [The] The best way to cut and paste this recipe is to squeeze the hot water into a small stirringe. Mix it with enough water to form a paste, then pour it into a pot.
+Sprinkle the sauce over your vegetables. Be careful not to peel and run the sauce in the oven to finalize the dish.
+Michele Pezza is an author, journalist, and editor who specializes in agricultural sustainability
+  >>> [Scientists have discovered] Scientists have discovered that nine of the 11 new species of beauties are found in Mexico. And that there are 135 new species of ants.
+The number of known species is estimated to be 140,000 to 150,000. But that number would be much higher if there were ten of them, or a smaller number if there were only a
+step 145200/250000 | loss 2.8234 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,118 tok/s | epoch 1
+step 145400/250000 | loss 2.8444 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,121 tok/s | epoch 1
+  >>> val_loss: 3.1373 | bpt: 4.5261 | true_bpb: 1.4560
+  >>> [The] The Thorntonsonian Library, 1973). The second collection focuses on the private collection of mass means, using the works of Albert W. Hirschfeld and Franz Schlegel. It is a collection of stepped-down maps, relatively large maps, and printing drawings illustrating the historic and social history of Montana. The collection contains maps of the Montana real world documents, a Turkish inscription, 1
+  >>> [Scientists have discovered] Scientists have discovered that the next computer simulation to look at the molecular molecular dynamics of a molecule will be very helpful in the design of future therapies for the devastating condition known as leukaemia.
+The relationship between the amount and the quality of a molecule is the basis of molecular dynamics studies. The molecule of interest has to be the best residues from the beginning of the molecule, before the end of the molecule is evaluated.
+step 145600/250000 | loss 2.8389 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,119 tok/s | epoch 1
+step 145800/250000 | loss 2.8417 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,122 tok/s | epoch 1
+step 146000/250000 | loss 2.8236 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,126 tok/s | epoch 1
+  >>> val_loss: 3.1360 | bpt: 4.5243 | true_bpb: 1.4554 *BEST*
+  >>> [The] The Camels of the Nigh was a series of conical radial cannons mounted on the northern pole of the drive main. The first were mounted on the small 33” 31” gal travails that were used to launch the return march. The other were mounted on the 31” 29” gal travails that were used to launch the return march. The latter half
+  >>> [Scientists have discovered] Scientists have discovered that the mineral oxide
+is a sedimentary rock and that its composition is a result of both both
+the chemical oxidation of the rock and its subsequent deposition in
+the mantle. The cyclography of the minerals is a useful tool for
+technicians to investigate giant magma sources in the mantle. Of
+course, it can be used for different analyses.
+There are many uses for aocl
+step 146200/250000 | loss 2.8159 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,119 tok/s | epoch 1
+step 146400/250000 | loss 2.8071 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,123 tok/s | epoch 1
+  >>> val_loss: 3.1375 | bpt: 4.5265 | true_bpb: 1.4561
+  >>> [The] The African-American community, especially in South Carolina, had been writing their own texts, mainly about their experiences and experiences in the country itself. While the emphasis on the African-American experience was aimed primarily at white readers, some African Americans were already writing in the context of the Civil Rights narrative. This paper explores how the 1955 March on Washington and its aftermath, as well as how this paper
+  >>> [Scientists have discovered] Scientists have discovered that the state of the universe in a supermassive black hole has a gravity that is 46.6% of its current value, and that its mass has doubled since the Big Bang. The results were published in the journal Nature.
+The observation was made using two optical telescopes located at the Chandra X-ray Observatory in Chile, located in Chile, and the Large Hadron Collider in
+step 146600/250000 | loss 2.8231 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,120 tok/s | epoch 1
+step 146800/250000 | loss 2.8125 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,124 tok/s | epoch 1
+step 147000/250000 | loss 2.8140 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,127 tok/s | epoch 1
+  >>> val_loss: 3.1354 | bpt: 4.5235 | true_bpb: 1.4552 *BEST*
+  >>> [The] The Granthanes Heritage Park begins on 15th June 2017. The park takes its name from Granthanes, a hill that remains a popular destination for visitors of the region. Named for its salt. It occupies an area of 16,781 hectares (approx. 50,921 sqft) and is famous for the historic Le Conte
+  >>> [Scientists have discovered] Scientists have discovered a new way for tracking communications, also known as a “branch-tracker” , which allows users to “bring their private money to a startup, using it to provide private companies and government agencies with a link to a real-time news source. The technology is a similar to the “head time” or “massive click” technology developed for Facebook, which was developed by U.S. company Bison
+step 147200/250000 | loss 2.8418 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,121 tok/s | epoch 1
+step 147400/250000 | loss 2.8171 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,124 tok/s | epoch 1
+  >>> val_loss: 3.1342 | bpt: 4.5217 | true_bpb: 1.4546 *BEST*
+  >>> [The] The Pennsylvania State Library (SPM) provides information on local history and culture of Philadelphia. There are many more interactive city maps and information on the history of Philadelphia. If you can’t find the information you are looking for, you can find it at www.ppsc.org.
+The Pennsylvania State Archives is one of several important pieces of correspondence relating to the Philadelphia and surrounding regions. The Pennsylvania State Archives has made
+  >>> [Scientists have discovered] Scientists have discovered a new mechanism that could be used to help people of all ages become physically fit and match their unique personality and mental capacity.
+The study was published in the journal Brain Research.
+- The scientists are using machine learning with genetic data to learn about the genetic abnormalities that may be contributing to the disorder.
+- The researchers were able to pinpoint which genes affected a number of diabetes-related disorders.
+- The researchers then
+step 147600/250000 | loss 2.8551 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,118 tok/s | epoch 1
+step 147800/250000 | loss 2.8239 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,122 tok/s | epoch 1
+step 148000/250000 | loss 2.8290 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,125 tok/s | epoch 1
+  >>> val_loss: 3.1362 | bpt: 4.5246 | true_bpb: 1.4555
+  >>> [The] The Sasakhi are a third order of the Sankhars, or little gods. They are most often associated with numerical figures, not only the spirit of a god, but also the god himself. The calculation of the number of the Sankhars with regard to a given number is directly related to that number, but the sage is never limited. In fact, the numbers for all the
+  >>> [Scientists have discovered] Scientists have discovered that in a new study, scientists have identified a key gene led to how plants appropriately respond to drought.
+“Biological diversity is very important in the adaptation of plants to drought,” said Thomas Blunden, a professor and chair of the Botany Department’s Department of Plant Sciences and a professor at the University of Cincinnati who was not involved in the study.
+“Human diversity is a very important part of
+step 148200/250000 | loss 2.8308 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,123 tok/s | epoch 1
+step 148400/250000 | loss 2.8434 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,126 tok/s | epoch 1
+  >>> val_loss: 3.1357 | bpt: 4.5239 | true_bpb: 1.4553
+  >>> [The] The irony of this approach is its failure to adequately capture the complex and complex dynamics of human behaviors. Two years into the project, the team has uncovered a complex and complex system for the production of the human brain. The team’s knowledge of the brain’s function and its health characteristics as well as the way in which the brain functions are key questions in understanding whether the human brain doesn’t function properly. This
+  >>> [Scientists have discovered] Scientists have discovered mechanisms that could be explained by the special effects of the planet's violent magnetic fields.
+As a result, the astronomers were able to determine that the planets' magnetic fields are significantly different than expected, indicating that the planets are not the same as they appear to be.
+David Schmidt, a member of the research team working on the new study, said: "The planets are so different from the rest of
+step 148600/250000 | loss 2.8132 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,124 tok/s | epoch 1
+step 148800/250000 | loss 2.8175 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,127 tok/s | epoch 1
+step 149000/250000 | loss 2.8083 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,130 tok/s | epoch 1
+  >>> val_loss: 3.1364 | bpt: 4.5249 | true_bpb: 1.4556
+  >>> [The] The Golgi is the largest and deepest organ in the body. This organ uses the movements in the kidneys to move the blood away from the body. The arteries, veins, and arteries are important organs that move blood throughout your body.
+The Golgi is the main organ that distributes blood throughout your body. It carries oxygen and other necessary nutrients into and from your cells. It also has a large supply of
+  >>> [Scientists have discovered] Scientists have discovered that the human brain is among the largest and most complex systems that ever evolved. In human brain, the brain and spinal cord have approximately 72,000 neurons, together with nerve control centers and brain tissue considered the brain and spinal cord. The differences between the human brain and the spinal cord are startlingly similar.
+The differences between the human brain and the spinal cord are strikingly similar
+step 149200/250000 | loss 2.8258 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,128 tok/s | epoch 1
+step 149400/250000 | loss 2.8193 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,131 tok/s | epoch 1
+  >>> val_loss: 3.1371 | bpt: 4.5259 | true_bpb: 1.4559
+  >>> [The] Theoretical physicist would tell you that the only way to predict the generalisation of a complex system is you can predict an entire system. With your intuition, then empiricist, or at least liberal, present arguments.
+Johannesburg’s implication is that numerous attempts to identify the underlying causal relationships between complex systems (as, say, the generalized energy theory of relativity) fail. A recent article
+  >>> [Scientists have discovered] Scientists have discovered a very strong network of connection between all of the different systems within the universe. Such an extensive network of interstellar currents is constantly collapsing into very small black holes. The phenomena is referred to as "Rapidly operating Super-Ears," or RFOBs.
+The most important factor that is responsible for the expansion of the universe is neutron stars, tiny black holes thought to reside within the Milky Way
+step 149600/250000 | loss 2.8411 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,129 tok/s | epoch 1
+step 149800/250000 | loss 2.8293 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,133 tok/s | epoch 1
+step 150000/250000 | loss 2.8455 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,136 tok/s | epoch 1
+  >>> val_loss: 3.1361 | bpt: 4.5245 | true_bpb: 1.4555
+  >>> [The] The Fulcheron region, an area of the United States located near the Texas border, is home to a myriad of different types of orchids, including orchons, stover, and even larches. All are very similar to orchids but a few varieties are larger, sometimes growing to 15 feet in diameter. The oldest trees are 12 to 15 feet tall.
+  >>> [Scientists have discovered] Scientists have discovered evidence that plants are up to 350,000 years old, and since the time of the dinosaurs, these fossils have been found in many of the same habitats. In the past, humans have also been thought to be ‘up to 100,000 years’ old, and hence they have been found in a number of places, the most recent being near
+step 150200/250000 | loss 2.8216 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,132 tok/s | epoch 1
+step 150400/250000 | loss 2.8338 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,135 tok/s | epoch 1
+  >>> val_loss: 3.1368 | bpt: 4.5254 | true_bpb: 1.4558
+  >>> [The] The Hofstra Law Foundation’s Legal Intervention Tool provided an overview of the law in terms of ‘legal standards and procedures’ that define the consequences civil servants’ actions. Per the Law of the Council of Canada, the Law of the Council of Canada defines the legal requirements that cells and other entities must follow while accepting their client’s consent to a procedure. The guideline is also referred to as ‘the Law of
+  >>> [Scientists have discovered] Scientists have discovered that the study showed that the people in the study could aid in saving 1000 people’s life.
+The findings are of great importance for the heart style. The study also explains that the researchers said that their study showed that the people in the study could help in saving 1000 people’s life.
+According to the study, the results of the study are not as alarming as
+step 150600/250000 | loss 2.8127 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,133 tok/s | epoch 1
+step 150800/250000 | loss 2.8175 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,136 tok/s | epoch 1
+step 151000/250000 | loss 2.8309 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,140 tok/s | epoch 1
+  >>> val_loss: 3.1365 | bpt: 4.5251 | true_bpb: 1.4557
+  >>> [The] The Olympic Games in 1900, a country in the midst of an extensive cultural exchange, had a population of 1.1 billion. About 45,000 of them were American, some of them Spaniards. According to World Bank statistics, an estimated 90 percent of the total population of the European countries lived in the Low Countries, a colony of the Russian
+  >>> [Scientists have discovered] Scientists have discovered that certain areas of the body are more apt to fuse at the age of 12 than others. This comes about due to the ability of certain genes to fuse or fuse with specific regions of the body. This process is also called leucinogenesis. At the same time, certain cells of the body can fuse with specific genetic regions of the body. In this way, the cell can fuse
+step 151200/250000 | loss 2.8249 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,137 tok/s | epoch 1
+step 151400/250000 | loss 2.8218 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,141 tok/s | epoch 1
+  >>> val_loss: 3.1346 | bpt: 4.5223 | true_bpb: 1.4548
+  >>> [The] The SherbKey Project
+The SherbKey Project (or Wright Hill) was a program that helped cover the cost of the solid chemicals needed for the production of metal and steel products from the 1920s through the 1940s. It was a crash program that came to be known as the SherbKey Project. The SherbKey Project was the first massive American
+  >>> [Scientists have discovered] Scientists have discovered that healthy proteins on our skin are being damaged by certain chemicals used in skincare and cosmetics. Here are a few ways you can use skincare products to help keep them looking and feeling better.
+- Vitamin D: Vitamin D is important for bone health, helps to strengthen bones and teeth, and helps to reduce redness and redness. One study published in the American Academy of Dermatology found that patients with
+step 151600/250000 | loss 2.8459 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,138 tok/s | epoch 1
+step 151800/250000 | loss 2.8042 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,142 tok/s | epoch 1
+step 152000/250000 | loss 2.8328 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,145 tok/s | epoch 1
+  >>> val_loss: 3.1348 | bpt: 4.5226 | true_bpb: 1.4549
+  >>> [The] The work was performed by a group of fourth graders, including some who will soon be enrolled in one of the world’s largest high school science programs.
+The school district’s 2,000 students are in a competitive competition to choose the best teachers of their choice. The students studied the most complicated geological time periods in Earth history, such as our own.
+The science teacher was so impressed and
+  >>> [Scientists have discovered] Scientists have discovered that there are some 10,000 strain mutations each year that can affect thousands of genes. Their study suggests a potential drug that could speed up the-sweeping procedure.
+The scientists analyzed a set of 13 genes and compared them to flu strains from people diagnosed with the H1N1 flu.
+The scientists first tested the H1N1 strain in cholera-causing
+step 152200/250000 | loss 2.8288 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,142 tok/s | epoch 1
+step 152400/250000 | loss 2.8327 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,146 tok/s | epoch 1
+  >>> val_loss: 3.1332 | bpt: 4.5203 | true_bpb: 1.4541 *BEST*
+  >>> [The] The Vedas' version of this sixth spiritual text, the Mahabharata, is the oldest surviving version to date, dating more up to 3,700 years ago than is usually believed. The Mahabharata (by classical authors and Vedanta scholars) was much shorter in length than its predecessor, and included many vivid, sometimes sometimes jarring passages. It offers a fuller picture
+  >>> [Scientists have discovered] Scientists have discovered that a certain type of bacteria is responsible for the growth of mold on the underside of a window sill. Scientific studies have also revealed that a certain type of mold is responsible for mold growth on the inside. The scientists discovered that one type of mold, known as a spores, produces spores that are not paint-resistant. The spores must be heated to extremely high temperatures to die off.
+The
+step 152600/250000 | loss 2.8065 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,140 tok/s | epoch 1
+step 152800/250000 | loss 2.8117 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,143 tok/s | epoch 1
+step 153000/250000 | loss 2.7917 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,146 tok/s | epoch 1
+  >>> val_loss: 3.1348 | bpt: 4.5226 | true_bpb: 1.4549
+  >>> [The] The present study will focus on the three-dimensional (3D) determination of structure and immateriality of the life-relationships of the occurrence of the human skeletal muscle. Furthermore, the methods will be discussed with applications to human physiology, physiology of the human (e.g., non-human primates, rodents, humans, dogs), and to environmental and social phenomena that influence human health. The human skeletal muscle
+  >>> [Scientists have discovered] Scientists have discovered that the night sky is a great place to look for stars. The reason why we get a lot of is because there are lots of the stars that make up the night sky.
+Even though you can’t see starlight, you can see at night, either as a bright cloud of light or as a dark cloud of gas and dust. If you’re in a bright place, such as the sky
+step 153200/250000 | loss 2.8229 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,144 tok/s | epoch 1
+step 153400/250000 | loss 2.8404 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,147 tok/s | epoch 1
+  >>> val_loss: 3.1350 | bpt: 4.5229 | true_bpb: 1.4550
+  >>> [The] The Gates of the Sea: Affordances, Conflicts, and the Submarine Warfare of the First World War (1993) told of how European submarines fought against a variety of Allied vessels, including Germany's massive shipyards and ships of the Western and Eastern Seas. In 1948, the German Government created an intelligence advisory committee to investigate the common naval concerns of Germany.
+  >>> [Scientists have discovered] Scientists have discovered that some parts of a new species' genetic code (mutations) can be adapted to exploit changing climate.
+"Backgrounds on adaptation will shed light on how the Bornean’s use of the environment affects nature’s response to climate change in general," co-author Dr. Bariya Daula Dhakal of the University of Francisco in Spain noted.
+The Bornean is known for its
+step 153600/250000 | loss 2.8289 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,145 tok/s | epoch 1
+step 153800/250000 | loss 2.8327 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,148 tok/s | epoch 1
+step 154000/250000 | loss 2.8106 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,152 tok/s | epoch 1
+  >>> val_loss: 3.1345 | bpt: 4.5222 | true_bpb: 1.4547
+  >>> [The] The Capitute of Gods and Gods, is the Book of the Dead in India. It is also known as the Book of the Dead. It is a sacred sacred text relating to the movement of the human body to the underworld, the spiritual unit of the Hindu pantheon. The book has been prepared by the Vaishnava princess Mirunah who is the mother of Lord Rama. She
+  >>> [Scientists have discovered] Scientists have discovered that a gene that facilitates metabolise of glucose in the liver could be able to control glucose levels in the blood. This may affect blood sugar levels in people with diabetes, which may lead to the development of complications.
+Besides the increased risk of blood clots, the researchers say that the gene is also a potential source of new drugs to control blood sugar levels in people with diabetes.
+Cancer patients with
+step 154200/250000 | loss 2.8352 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,149 tok/s | epoch 1
+step 154400/250000 | loss 2.8297 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,153 tok/s | epoch 1
+  >>> val_loss: 3.1337 | bpt: 4.5209 | true_bpb: 1.4543
+  >>> [The] The crowd was out for him. His heart became deep with gratitude and admiration for the erudite and determined knight.
+
+The Jolly Maltese gathered around him, their hearts pounding with excitement and energy. They knew that they had discovered a beautiful and meaningful queen who had made a lasting legacy in their culture and history.<|im_end|>
+  >>> [Scientists have discovered] Scientists have discovered a new mechanism by which the brain redeposits memories and becomes reactivator together with the brain.
+The findings, published in the journal Nature, explain why Alzheimer’s disease is still a problem today.
+The research team discovered the mechanism by looking at the symptoms of Alzheimer’s disease and how to treat it.
+Symptoms of Alzheimer’s disease include memory problems, memory trouble and problems with cognition.
+The
+step 154600/250000 | loss 2.8154 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,150 tok/s | epoch 2
+step 154800/250000 | loss 2.8054 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,153 tok/s | epoch 2
+step 155000/250000 | loss 2.8086 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,156 tok/s | epoch 2
+  >>> val_loss: 3.1323 | bpt: 4.5189 | true_bpb: 1.4537 *BEST*
+  >>> [The] The Gates of JudgementThe Gates of Judgement
+The Gates of Judgement
+The Gates of Judgement
+The Gates of Judgement, The Gates of Judgement, The Gates of Judgement, The Gates of Judgement, The Gates of Judgement (or The Gates of Judgement in Hebrew or the Gates of Judgement and Judgement in Aramaic), The Gates of Judgement (
+  >>> [Scientists have discovered] Scientists have discovered changes in the way the brain converts bone density, the brain’s elasticity and the ability of neurons to communicate. This is part of the brain’s ability to coordinate movements and communicate with other brain cells. The researchers said it may help explain why brain abnormality or disruptions can occur in such a way.
+Scientists believe the new study demonstrates the importance of dementia as a cause of dementia in the brain and suggests
+step 155200/250000 | loss 2.8145 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,148 tok/s | epoch 2
+step 155400/250000 | loss 2.7961 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,152 tok/s | epoch 2
+  >>> val_loss: 3.1332 | bpt: 4.5203 | true_bpb: 1.4541
+  >>> [The] The Sabatubo uprising began in 1974. In 2000, the army of Streppuou entered into an agreement with the Serb army to establish a U.S. Army unit. However, by 2004, the army had already ceased to exist. In March 2005, the Serb troops began a reintegration into Sabat
+  >>> [Scientists have discovered] Scientists have discovered a gene that blocks the ability of the brain to absorb sleepless night sleep. Now, researchers have found a way to stimulate this bypass. The new work puts a new spin on the common theory that sleeplessness means sleep deprivation is linked to sleep problems.
+The paper describes this research in the journal Proceedings of the National Academy of Sciences. The findings, which can appear online today (Nov. 4
+step 155600/250000 | loss 2.8047 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,149 tok/s | epoch 2
+step 155800/250000 | loss 2.7907 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,152 tok/s | epoch 2
+step 156000/250000 | loss 2.8200 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,156 tok/s | epoch 2
+  >>> val_loss: 3.1328 | bpt: 4.5197 | true_bpb: 1.4539
+  >>> [The] The Casiboard TB Mk II, the first tank carrier to be bought by the USA due to the shortage of main deck power, was the first tank carrier to be purchased by the USA. The Special recently acquired Pan Am Air Force had only 20000 direct-line tanks in the US Navy. In 1963, the Rompans had one 17-gun gun
+  >>> [Scientists have discovered] Scientists have discovered that it is crucial to find and develop the right endocrinologist to deliver the right level of medication to the patient with all the right drugs to relieve his or her symptoms.
+It is essential to protect the patient from harm and make him or her feel healthy in all the ways. The body can achieve proper functioning in the body through the brain, and through regular rest. The patient must have good nutrition
+step 156200/250000 | loss 2.7939 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,153 tok/s | epoch 2
+step 156400/250000 | loss 2.8141 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,157 tok/s | epoch 2
+  >>> val_loss: 3.1324 | bpt: 4.5191 | true_bpb: 1.4538
+  >>> [The] The Mandelbrot Papers, 1930, pp. 31-32
+- A. E. Diegel and A. H. Steiner, The Story of American Literature, 1915, pp. 44–50
+- A. E. Diegel and G. H. Steiner, The Past and Present, 19
+  >>> [Scientists have discovered] Scientists have discovered a ratio of the height of two plants from the same species up to 10cm, 10cm apart, and 100cm between them. Their original discovery has finally been confirmed as if exactly 10cm and 10cm are the same species.
+John Estes, assistant professor at the University of Queensland’s School of Environment, School of the Environment, and
+step 156600/250000 | loss 2.7956 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,154 tok/s | epoch 2
+step 156800/250000 | loss 2.8288 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,158 tok/s | epoch 2
+step 157000/250000 | loss 2.8051 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,161 tok/s | epoch 2
+  >>> val_loss: 3.1320 | bpt: 4.5185 | true_bpb: 1.4536 *BEST*
+  >>> [The] The Camusian Temple
+The magnificent Ptolemy VII and his grandsons are now at the entrance to the second-largest shrine in the world. In later times, the temple was used as the temple of Amencheon the Great (the "Great" Ptolemy VII) and the Osiris I (the "Great").
+The Palaeoclimax was made by placing a large
+  >>> [Scientists have discovered] Scientists have discovered a genetic relationship between the Covid-19 strain and a milder form of the very long COVID-19 strain.
+“This is the first time we’ve been able to tell a person from the pathogen or strain,” said author and co-author of the study, Christina Machanos, assistant professor and immunology professor of immunology.
+The researchers used COVID-19 to replicate the Covid-
+step 157200/250000 | loss 2.8143 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,155 tok/s | epoch 2
+step 157400/250000 | loss 2.8093 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,158 tok/s | epoch 2
+  >>> val_loss: 3.1327 | bpt: 4.5195 | true_bpb: 1.4539
+  >>> [The] The sagus was a fourth layer of muscle tissue (caudal lunate fibrospinal), that was all about the lungs.
+The sagus had a rich vascular network that fed the lungs and blood vessels.
+It also had many layers of tissue, including the connective tissue (tissue that lines the skin and joints), tendons (coal)->he fibrous, connective tissue, and
+  >>> [Scientists have discovered] Scientists have discovered that they have different strains of bacteria that can cause the illnesses such as candida and methicillin-resistant Staphylococcus aureus, or MRSA, to develop in the general population.
+The team of researchers at the University of Washington worked with students in the United Kingdom to learn more about the microbes and how the microbes can affect the body.
+The findings were published in the Journal of Clinical
+step 157600/250000 | loss 2.8108 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,156 tok/s | epoch 2
+step 157800/250000 | loss 2.8024 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,159 tok/s | epoch 2
+step 158000/250000 | loss 2.8034 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,162 tok/s | epoch 2
+  >>> val_loss: 3.1316 | bpt: 4.5180 | true_bpb: 1.4534 *BEST*
+  >>> [The] The Hoistman at 1940-1941, which was a solitary military camp in northeastern Portugal, was one of the first places in Europe to combat a strong nationalist and anti-nationalist movement. Inspired by the campaign generated by the raids on the Grenadiers, the Hoistman at 1940-1941 was one of only two camps to
+  >>> [Scientists have discovered] Scientists have discovered one of the first functions of air in the body, which is the movement of oxygen through the blood. As a result, the body is actually able to perform the functions without the assistance of the lungs. In addition to the respiratory function and respiratory function, breathing also plays the role in helping the body to absorb minerals and other substances. Respiratory functions include the clearance of the blood and initiating the res
+step 158200/250000 | loss 2.8012 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,156 tok/s | epoch 2
+step 158400/250000 | loss 2.8040 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,159 tok/s | epoch 2
+  >>> val_loss: 3.1320 | bpt: 4.5185 | true_bpb: 1.4536
+  >>> [The] The Gullah in India is of Kashmiri origin. The Gullah culture also includes modern Hindi. The Gullah people  are the buffalo who are the ancestors of the Mughal Empire. The Gullahs are descended from the Mughal Empire. The Mughal Empire was the dominant power in India during the 17th to 19th century. The British took over the
+  >>> [Scientists have discovered] Scientists have discovered that the bacteria can cause certain types of cancer, such as esophageal cancer. So they are offering suggested treatments for patients with these cancers, and it will be conducted in a clinical research facility in the next two years.
+The researchers think that more research on the bacteria will be able to be conducted in the future. The study was published in the journal Cancer Research.
+For further information, Dr.
+step 158600/250000 | loss 2.8104 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,157 tok/s | epoch 2
+step 158800/250000 | loss 2.8111 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,160 tok/s | epoch 2
+step 159000/250000 | loss 2.7970 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,164 tok/s | epoch 2
+  >>> val_loss: 3.1316 | bpt: 4.5180 | true_bpb: 1.4534
+  >>> [The] The Beginner was a kind of bird that had no bird-like features, much like in the books. If you think you should be an amateur bird watching, you can take the "Roots" book and make a bird trunk out of your choice of trees. The books are scheduled to be available in 2014. You can get them quite easily at the book store. You can find a
+  >>> [Scientists have discovered] Scientists have discovered that there's an immense amount of water on the surface of Mars, which is at least as salty as water on Earth. Even though there are only a few hundred thousand miles of water on Mars, the researchers think it may well make use of it as a chance to grow up on other planets like the Earth. (And instead of hibernating in the colder, dry desert of Mars, astronauts
+step 159200/250000 | loss 2.8131 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,161 tok/s | epoch 2
+step 159400/250000 | loss 2.8174 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,164 tok/s | epoch 2
+  >>> val_loss: 3.1332 | bpt: 4.5202 | true_bpb: 1.4541
+  >>> [The] The Marshall Houlsen creative philosophy sought to unify the work of a variety of artistic movements with a focus on space. This philosophy, the name is often used to emphasize the use of physical and spatial relationships. Schopenhauer acknowledges that "from time immemorial, the use of space has been felt as the most beautiful form of behavior" (Walden, 34). Schopenhauer describes
+  >>> [Scientists have discovered] Scientists have discovered a unique mechanism that binds to a gene involved in the development of the adult red blood cell. This is the key to the development of the first fully functional blood cell, as it may help to regulate the activity of the immune system. The findings are published in the online edition of Science. The team was led by Joachim Garun of the Moshe Institute for Molecular and Cellular Biology, Jewish
+step 159600/250000 | loss 2.8076 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,162 tok/s | epoch 2
+step 159800/250000 | loss 2.7945 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,165 tok/s | epoch 2
+step 160000/250000 | loss 2.8089 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,168 tok/s | epoch 2
+  >>> val_loss: 3.1322 | bpt: 4.5188 | true_bpb: 1.4537
+  >>> [The] The giraffe is a rare sight in the Sierra. As you go along you will find the dodo, which has a thick, dark coat and a short tail with a long tail. In addition to its distinctive name, you can also find the tiger, which is a bit more familiar. The mountain lion is a very rare sight in the Sierra.You can use a longitudinal truck that may only be
+  >>> [Scientists have discovered] Scientists have discovered two new genes that are critical for cell growth. The cells in which they are found have been mutated so that they cannot properly process oxygen and nutrients. The new gene has been identified as an important driver of cell death in mice. The study findings were published in the December 2004 Journal of Neurology.
+"Our results reveal one important mechanism which helps cells to enter tissue and to reside,"
+step 160200/250000 | loss 2.7967 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,165 tok/s | epoch 2
+step 160400/250000 | loss 2.8215 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,168 tok/s | epoch 2
+  >>> val_loss: 3.1309 | bpt: 4.5170 | true_bpb: 1.4531 *BEST*
+  >>> [The] The Vandals invaded Spain in 1211. This conflict led to a major British victory. The arrival of the Pope in 1216 was the first formal Battle of the Realm, a result of determination by Pope Benedict the Great to establish a unified Holy Roman Empire.
+The Normans appealed to the New World powers in order to encourage each other to rule their Barbarian lands in the
+  >>> [Scientists have discovered] Scientists have discovered that a molecule that existing in the brain can change when exposed to drugs that are used to treat certain cancer therapies, including Oncotherapy, which is a class of compounds designed to block the effects of cancer drugs. A group of investigators led by Virginia Tech neurosurgeon Michael Thoot, PhD, based on work in the lab of Amyloid lupus at the University of Washington, report their findings
+step 160600/250000 | loss 2.8034 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,162 tok/s | epoch 2
+step 160800/250000 | loss 2.8196 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,166 tok/s | epoch 2
+step 161000/250000 | loss 2.8067 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,169 tok/s | epoch 2
+  >>> val_loss: 3.1308 | bpt: 4.5167 | true_bpb: 1.4530 *BEST*
+  >>> [The] The enemy is all around you,
+Nor do I deny the advantage of their help.
+Troops have been out of my hands for many years, but if you are an American, you have come to believe that with the American people, you can take over the world with you, with the nation, with your own hands. As long as you maintain your own self-confidence, you can lead the way.
+  >>> [Scientists have discovered] Scientists have discovered hundreds of new genes in the human genome that could potentially be used to stop disease. Researchers are already monitoring them using a variety of methods, including mapping genes down to gene expression. One of the biggest challenges is that the process requires small amounts of in vitro DNA, but it is possible to create their own. Most of the genome is still unclear, but most of it can be targeted by nanoparticles.
+step 161200/250000 | loss 2.7989 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,163 tok/s | epoch 2
+step 161400/250000 | loss 2.8160 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,166 tok/s | epoch 2
+  >>> val_loss: 3.1303 | bpt: 4.5160 | true_bpb: 1.4527 *BEST*
+  >>> [The] The Bartons were created in the early 1800s to imitate the fish in Africa. They evolved in response to environmental conditions. During the late 1800s, the Bartons became increasingly endangered due to their overexploitation for their meat and their fins, which were marked worldwide.
+Some species in the Barton family have been introduced into the wild, such as the European
+  >>> [Scientists have discovered] Scientists have discovered that while there are many different types of viruses, the most common types of viruses are the most common ones. This is because viruses have similarities to other species and have different cell structures. Our immune system uses a process called cell-mediated immunity to achieve this.
+Adapted from the book:
+Biologic Immunity for Humans by Ivan Pawlaczka (1994)
+Author: Ivan Paw
+step 161600/250000 | loss 2.8136 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,160 tok/s | epoch 2
+step 161800/250000 | loss 2.8181 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,164 tok/s | epoch 2
+step 162000/250000 | loss 2.7814 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,167 tok/s | epoch 2
+  >>> val_loss: 3.1298 | bpt: 4.5153 | true_bpb: 1.4525 *BEST*
+  >>> [The] The dichochromatoid diet is a low relative fat diet that is very low in both fat and protein, but rapidly delivers fat to your body’s cells. This diet is a healthy one, and it is a major electrolyte rich food, thus it is not a bad option for people who are only 1 to 2 years old. The high protein content of this diet is often associated with
+  >>> [Scientists have discovered] Scientists have discovered that the international team’s research vessel onboard EVE, the Galileo, visited a polar region of the planet — the largest known in the solar system — at an altitude of 110,000 kilometers (60,000 miles) below the Earth’s surface.
+The mission began in January 2004, when the ship sailed from the Arctic’s southern shore and
+step 162200/250000 | loss 2.7912 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,161 tok/s | epoch 2
+step 162400/250000 | loss 2.8261 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,164 tok/s | epoch 2
+  >>> val_loss: 3.1299 | bpt: 4.5155 | true_bpb: 1.4526
+  >>> [The] The Triumph of the City is the story of the five-year effort by the Italian police to put up the Communist-era structure in the city. It was not, however, the work of the leadership, nor as much the work of the military commanders, as it is the work of the masses who decided to see the building finally laid out.
+The central figure of the story is a young rebel leader of the
+  >>> [Scientists have discovered] Scientists have discovered another protein that may help protect against both breast and testicular cancers.
+Around 10 million Americans are treated for breast cancer each year, and 1,000 die from it. So, what if breast cancer could be prevented? The answer is the same for women, and it turns out that fat is one of the most important things to keep in check.
+There’s a very good
+step 162600/250000 | loss 2.7997 | lr 8.00e-04 emb 4.00e-04 | 667ms/step | 49,162 tok/s | epoch 2
+step 162800/250000 | loss 2.8365 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,165 tok/s | epoch 2
+step 163000/250000 | loss 2.8205 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,168 tok/s | epoch 2
+  >>> val_loss: 3.1298 | bpt: 4.5154 | true_bpb: 1.4526
+  >>> [The] The General Court decisions on the right to preserve the environment and the right to property in the absence of political opposition. The court rejected the case in which the Court held the lawfully protected operation would be same as applying the law to a private company in the state of Georgia.
+ESRI's legal office is located on 18th Street in Tallahassee, Florida.The use of photo filters in
+  >>> [Scientists have discovered] Scientists have discovered three genes wreaking havoc with the gut’s unique gut flora.
+A healthy gut may have small but significant impact on your mood, weight, and metabolism. It also helps to keep your digestive system healthy and eliminator of food particles and toxins in the gut. It also helps prevent a variety of diseases in the body, including diabetes and heart disease.
+Few of the significant gut issues have been well documented
+step 163200/250000 | loss 2.7973 | lr 8.00e-04 emb 4.00e-04 | 666ms/step | 49,165 tok/s | epoch 2