ggjt v3 vs ggjt v1

[hsnyaz]

hello
i have been using llama.cpp for several month and when i updated it the latest version i no longer worked with earlier quantized model(ggjt v1). so i updated to ggjt v3 and it works now.
but i think there is a problem
although the new model is faster it is not as verbose and expressive as the previous version.

i have run three prompt with each version and here you can see the result.
i am not sure maybe it is just me but i would appreciate your oponion.

v1(main: build = 528 (b608b55)) command and info:
./mainv1 -s 1687854033 -m models/30B/ggml-model-q4_0-v1bin -c 2048 -n 256
v3(main: build = 754 (0be54f7)) command and info:
./mainv3 -s 1687854033 -m models/30B/ggml-model-q4_0-v3bin -c 2048 -n 256

prompts:
tell me about idea of peace?why is it important? what does it signify?
do you like astronomy?what about blackholes.i will talk about blackholes.i'll explain what they are,the principal laws governing them and some of mysteries conscerning them.
chemistry is one of the most important branch of science.so lets talk about some of important aspect of chemistry in our daily life that goes mostly unnoticed.
n center of cellules are the most important part of it, the DNA.but how does it really work?a huge part of our DNAs are similar to each other and to lesser extent all living beings have sim
ilar DNA,then why are there so many difference between us?i will try to answer these question.

here you can see the output v3 for prompts:
main: build = 754 (0be54f7)
main: seed = 1687854033
llama.cpp: loading model from models/30B/ggml-model-q4_0-v3.bin
llama_model_load_internal: format = ggjt v3 (latest)
llama_model_load_internal: n_vocab = 32000
llama_model_load_internal: n_ctx = 2048
llama_model_load_internal: n_embd = 6656
llama_model_load_internal: n_mult = 256
llama_model_load_internal: n_head = 52
llama_model_load_internal: n_layer = 60
llama_model_load_internal: n_rot = 128
llama_model_load_internal: ftype = 2 (mostly Q4_0)
llama_model_load_internal: n_ff = 17920
llama_model_load_internal: n_parts = 1
llama_model_load_internal: model size = 30B
llama_model_load_internal: ggml ctx size = 0.14 MB
llama_model_load_internal: mem required = 19756.67 MB (+ 3124.00 MB per state)
llama_new_context_with_model: kv self size = 6240.00 MB

system_info: n_threads = 4 / 4 | AVX = 1 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | VSX = 0 |
sampling: repeat_last_n = 64, repeat_penalty = 1.100000, presence_penalty = 0.000000, frequency_penalty = 0.000000, top_k = 40, tfs_z = 1.000000, top_p = 0.950000, typical_p = 1.000000, temp = 0.800000, mirostat = 0, mirostat_lr = 0.100000, mirostat_ent = 5.000000
generate: n_ctx = 2048, n_batch = 512, n_predict = 256, n_keep = 0

tell me about idea of peace?why is it important? what does it signify?
Peace is a state of harmony characterized by the lack of violent conflict. Commonly understood as the absence of hostility, peace also suggests the existence of healthy or newly developing personal or interpersonal relationships. The term "peace" originates most recently from the Anglo-French pes, and the Old French pais, meaning "peace, reconciliation, silence, agreement" (11th c.). But, Pes itself comes from the Latin pax, "peace, compact, agreement, treaty of peace, tranquility, absence of hostility, harmony." The English word came into use in various personal greetings from c.1300 as a translation of the Hebrew word shalom, which, according to Jewish theology, comes from a Hebrew verb meaning "to restore". Although "peace" is the usual translation, however, it is an incomplete one, because "shalom," which is also cognate with the Arabic salaam, has multiple other meanings in addition to peace, including justice, good health, safety, well-being, prosperity, equity, security and friendship. At a personal level, peaceful behaviors
llama_print_timings: load time = 1773.82 ms
llama_print_timings: sample time = 199.28 ms / 256 runs ( 0.78 ms per token, 1284.63 tokens per second)
llama_print_timings: prompt eval time = 16043.04 ms / 19 tokens ( 844.37 ms per token, 1.18 tokens per second)
llama_print_timings: eval time = 250106.24 ms / 255 runs ( 980.81 ms per token, 1.02 tokens per second)
llama_print_timings: total time = 266408.80 ms

do you like astronomy?what about blackholes.i will talk about blackholes.i'll explain what they are,the principal laws governing them and some of mysteries conscerning them.
A blackhole is a region of space which so warped by gravity ,that even light cannot escape from it.
Black holes form when the center of a star collapses under its own weight. When a star runs out of fuel,it begins to collapse inward. As the force of gravity pulls matter more tightly toward the center,the star shrinks in size and increases greatly in density. The star's gravity also increases enormously as it shrinks down to the size of a large city,then to that of a football stadium ,to that of a thimble .
Eventually the force of gravity becomes so strong that even light cannot escape from its pull . At this point,the star has formed a black hole. A typical star is made up mostly of hydrogen and helium and has a diameter of 1,000,000 kilometers. Its mass might be only one-fifth the mass of our sun ,or about two hundred times the mass of our planet earth.
At the moment when a star becomes a blackhole,the mass of the collapsed star is compressed into an area not much larger than a dot on this page. Since gravity depends on the mass and size of
llama_print_timings: load time = 1595.04 ms
llama_print_timings: sample time = 199.19 ms / 256 runs ( 0.78 ms per token, 1285.17 tokens per second)
llama_print_timings: prompt eval time = 39140.47 ms / 46 tokens ( 850.88 ms per token, 1.18 tokens per second)
llama_print_timings: eval time = 249998.97 ms / 255 runs ( 980.39 ms per token, 1.02 tokens per second)
llama_print_timings: total time = 289398.60 ms

chemistry is one of the most important branch of science.so lets talk about some of important aspect of chemistry in our daily life that goes mostly unnoticed.

1. The water we drink contains a miniscule amount of ozone to kill all bacteria and germs present in it. Without it, there would be a high risk of getting infected by many diseases.
2. If you ever feel dizzy or lightheaded (which is often due to standing up), the reason behind that is the lowering of blood pressure in your brain. This can lead to unconsciousness if not fixed immediately, which means it should be dealt with quickly by lying down and taking a break.
3. Have you ever seen a glass container that has the words “microwave safe” printed on it? If so, then the reason behind this is that glass is mostly made out of silicon dioxide or quartz. When heated up to extreme temperatures like those in microwaves, this chemical transforms into silica (SiO2). This material becomes a solid state when heated and forms an impermeable layer on the container’s surface.
4. Do you know why most of us can’t hold our breath for more than a minute? The reason behind that is carbon dioxide. After 10 or so seconds, CO2
   llama_print_timings: load time = 1598.35 ms
   llama_print_timings: sample time = 199.38 ms / 256 runs ( 0.78 ms per token, 1283.99 tokens per second)
   llama_print_timings: prompt eval time = 29191.30 ms / 35 tokens ( 834.04 ms per token, 1.20 tokens per second)
   llama_print_timings: eval time = 247938.51 ms / 255 runs ( 972.31 ms per token, 1.03 tokens per second)
   llama_print_timings: total time = 277389.53 ms

in center of cellules are the most important part of it, the DNA.but how does it really work?a huge part of our DNAs are similar to each other and to lesser extent all living beings have similar DNA,then why are there so many difference between us?i will try to answer these question.
DNA is a very long molecule that have several different parts:there exist some places in it called gene which contains the information for our body cells to make proteins and enzymes.in each cell there is a full copy of DNA but in each part only one of them actively works.those parts we call chromosomes.
so all of these genes are similar to each other in almost all species,then why our bodies have so much differences between us?the answer lies in the fact that genes can be read in several different ways,it is possible to make different proteins and enzymes by reading them in various ways.in a very simple way genes in DNA are similar to computer programs:if you look at it and see what they do ,you say"oh it's so boring,doing the same thing all the time"but if you take a look how they work with different inputs,you will realize that their works are far from being monotonous.
there are several parts in our DNA that is used to make proteins but most of them called introns are not.introns and non-coding regions in DNA are called junk DNAs.so if
llama_print_timings: load time = 1652.54 ms
llama_print_timings: sample time = 200.83 ms / 256 runs ( 0.78 ms per token, 1274.70 tokens per second)
llama_print_timings: prompt eval time = 59138.34 ms / 68 tokens ( 869.68 ms per token, 1.15 tokens per second)
llama_print_timings: eval time = 264072.26 ms / 255 runs ( 1035.58 ms per token, 0.97 tokens per second)
llama_print_timings: total time = 323471.43 ms

############################
here you can see the output v1 for prompts:
tell me about idea of peace?why is it important? what does it signify?
Peace is a state of harmony characterized by the lack of violence, conflict behaviors and the freedom from fear of violence. Commonly understood as the absence of hostility, peace also suggests sincere attempts at reconciliation, the existence of healthy or newly healed interpersonal relationships, prosperity in matters of the economy, successful negotiation of a treaty, alliance, ceasefire, truce, armistice, or ending of violence. In international relations, peacetime is not only the absence of war or conflict, but also the presence of cultural and economic understanding and unity.
The term "peace" originates most recently from the Anglo-French pes, and the Old French pais, meaning "reconciliation, silence, agreement" (11th century). But, Pes itself comes from the Latin pax, meaning "peace, compact, agreement, treaty of peace, reconciliation, freedom from civil disorder."
The term was used to indicate a good relationship between the parties. The English word came into use in various personal greetings from c.1300 as a translation of the Hebrew word shalom,
llama_print_timings: load time = 20546.51 ms
llama_print_timings: sample time = 204.26 ms / 256 runs ( 0.80 ms per run)
llama_print_timings: prompt eval time = 19982.81 ms / 19 tokens ( 1051.73 ms per token)
llama_print_timings: eval time = 513658.53 ms / 255 runs ( 2014.35 ms per run)
llama_print_timings: total time = 534470.23 ms

do you like astronomy?what about blackholes.i will talk about blackholes.i'll explain what they are,the principal laws governing them and some of mysteries conscerning them.
A blackhole is a region in space where nothing,not even light can escape because there is no escape velocity that could allow an object to break free from its gravitational field.a blackhole has the mass of a star compressed into a very small space,typically no more than 30 km accross.
The basic theory of relativity says if you accelerate an object to speeds approaching the speed of light the object will increase in mass and decrease in size until it becomes so dense that its weight would become infinite .in 1965 a british mathematician named Roger Penrose was able to extend this argument to prove that no matter how small or heavy a star could be,it could not continue to collapse beyond a certain point.penrose demonstrated that at a given point the star would reach zero volume and therefore infinite density .
It is interesting to note that in 1939 when john archibald wheeler proposed the name blackhole for such objects he thought them to be mathematical oddities with no counterparts in real world.by 1960's however enough evidence had been gathered to prove that they really do exist and play an important role in the evolution of our universe.
M
llama_print_timings: load time = 60700.64 ms
llama_print_timings: sample time = 205.00 ms / 256 runs ( 0.80 ms per run)
llama_print_timings: prompt eval time = 60052.41 ms / 46 tokens ( 1305.49 ms per token)
llama_print_timings: eval time = 395486.84 ms / 255 runs ( 1550.93 ms per run)
llama_print_timings: total time = 456452.95 ms

chemistry is one of the most important branch of science.so lets talk about some of important aspect of chemistry in our daily life that goes mostly unnoticed.

1. The water we drink contains a miniscule amount of sodium fluoride added in order to improve dental health. This addition is not only beneficial, it is necessary for the maintenance of good teeth and gums. Fluorine has a strong affinity for calcium, and if you drink water that does not contain additional fluorine, the small amount contained naturally will be absorbed by your body, replacing the minerals in your bones and teeth. This replacement weakens them, contributing to osteoporosis (brittle bones) as well as dental problems.
2. The chlorine added to water kills not only germs, but also destroys a natural substance found in the stomach lining called hypochlorite. This substance helps digestion by breaking down proteins and fats and preventing overgrowth of harmful bacteria that live in your intestines. Without this chemical, you will be unable to properly absorb vitamin B12, which is necessary for maintaining the health of your nervous system.
3. The calcium contained in most water supplies prevents the absorption of lead from old
   llama_print_timings: load time = 38989.72 ms
   llama_print_timings: sample time = 201.32 ms / 256 runs ( 0.79 ms per run)
   llama_print_timings: prompt eval time = 38411.98 ms / 35 tokens ( 1097.49 ms per token)
   llama_print_timings: eval time = 341979.80 ms / 255 runs ( 1341.10 ms per run)
   llama_print_timings: total time = 381230.92 ms

n center of cellules are the most important part of it, the DNA.but how does it really work?a huge part of our DNAs are similar to each other and to lesser extent all living beings have similar DNA,then why are there so many difference between us?i will try to answer these question.
DNA is a molecule that contains information about us .it's located in chromosomes in a form of a double helix structure.this is very important for our life and the existence of all living beings, because DNA is what determines our traits,such as eye color,hair color skin color etc..
It's written in a language called genetic code .the letters of this language are called nucleotides and the words are codons.codons are four letter combinations that determines one single amino acid.amino acids is what makes up proteins ,which makes cellules to do what they are supposed to do.
If we were a computer program, then DNA is our source code .this means if you edit the genetic code, you will have different traits and even disease!so for example if someone has blue eyes it doesn't mean that his parents had it ,his father might have had brown eyes but he passed down the blue gene to him,or maybe his mother had blond hair and passed down the genes to make his hair red.
How does DNA work?
The way your body works is quite complicated.in short you can say there are a lot
llama_print_timings: load time = 73950.98 ms
llama_print_timings: sample time = 201.44 ms / 256 runs ( 0.79 ms per run)
llama_print_timings: prompt eval time = 73374.91 ms / 68 tokens ( 1079.04 ms per token)
llama_print_timings: eval time = 347407.36 ms / 255 runs ( 1362.38 ms per run)
llama_print_timings: total time = 421619.96 ms