Speech:Spring 2017 Jeffrey Gancarz Log
- Information - General Project Information
- Experiments - List of speech experiments
Week Ending February 7th, 2017
2/2: Went onto the Wiki page for the 2016 Spring Semester. I decided to go into the tools group and look around to see what their tasks were, what their goals were, and what sorts of software they were planning on implementing at the beginning of the semester and what actually ended up getting implemented.
2/3: Read through the current semester logs to see how others are doing, what they have accomplished, and if they have any concerns.
2/5: Download VPN software that is necessary to log in to the ceasar server from remote access. Once I download the software, I will want to make sure I can login to caesar from my remote laptop.
2/7: Add an experiment to the experiments directory and to the wiki page and run the first train.
2/2: Learned that they looked into the software that was previously in the system already as well. They were able to upgrade some software, such as Sphinx Decoder, but other softwares were either all up to date or the previous Tools group did not think that they needed to upgrade the software, such as CMU Dictionary or Sphinx Trainer.
2/3: Realizing that nobody has really written any logs yet. There have been a couple people who have started, but overall people have not started writing logs. I'm assuming they are going to start soon considering we have to do a mandatory 4 logs per week.
2/5: Was able to download the software that is required for remote access and was also able to change my password.
2/7: I could not figure out as of now how to run my first experiment. I have spent hours trying to talk to other students over the past few days, have read up on other peoples logs, and have tried everything. I continue to get an error message when running the addExp.pl script. It's getting frustrating that I can't figure it out when every other person has been able to do it successfully.
2/2: I definitely would like to do some research into what these software's' capabilities are. In order to figure out what may need to be added or upgraded this semester, it is important that I research the software already installed in the system and figure out the functionalities of each.
2/3: Continue to do some research into what these software capabilities are. Continue reading and looking to see if we need to or could upgrade any current software.
2/5: Talk to the other members of my group and continue working on the draft of the proposal. Continue researching some of the software we talked about in class as a group, such as Torque, MLLT modeling, and Pocket Sphinx to learn about them and add sections to the proposal.
2/7: continue to read other logs and the wiki page to see if I can figure out how to create my first experiment. If I cannot figure it out I will most likely have to get help from other students in the classroom but hopefully it doesn't come down to that and I can figure it out before hand.
2/2: I do not know a lot about the capstone project in a whole or any types of software that the project currently uses. I also do not know much about what the Tools groups of previous semesters have done to make the project better. It will be my responsibility to do as much reading and research as possible to learn the project as soon as possible and determine software that may benefit the project.
2/3: Same concerns as of 2/2.
2/5: No new concerns at this point. Basically still trying to figure everything out within the project and make sure our group is on the same page for what we want to write about for our groups draft of the proposal.
2/7: the directions for the addExp.pl are not the best. I have been having to jump between multiple pages and many messages from students on how to create my first experiment and run my first train. Its frustrating to put this much time and effort into something and not see the results I am looking for.
Week Ending February 14, 2017
2/8: I was able to figure out how to set up an experiment yesterday, however, I was not able to run a train. I want to continue with the "Run Train Setup Script" from the wiki page. It is important for me to figure this out so that I can not only have the correct files in the system and on the wiki page, but so that I can help with my team with running future experiments.
2/10: Log in and read logs of current semester students as well as the 2016 Spring Tools Group. While looking at the logs of the 2016 Tools Group, I want to see exactly their research on Sphinx 3 so that I will be able to compare Sphinx 3 with Pocket Sphinx. I also was told that the tools group had implemented GCC onto caesar and obelix. Since we are going to be looking into G++, I think it would be wise to see how the installation process for GCC went and what problems the tools group had last year.
2/13: Read through my group members logs and see how they are doing with their research into both Pocket Sphinx and G++ compiler. Hopefully I will be able to see if they have learned anything or have any concerns that I will be able to help them out with. Otherwise, I can get a good understanding of what they have been researching and I can try to further what they have done or research another topic afterwards.
2/14: Continue researching GCC and G++. G++ is our main priority at this point. It would be helpful to have it installed on one of the drones in the next few weeks if possible. This will allow other groups to use the compiler in their work if they need to use G++ as well.
2/8: I was able to figure out the directions and set up the experiment folders in caesar as well as on the wiki page. I continued with the steps on the "Run Train Setup Script" and was able to go through the remaining steps and run the training. I am not sure if it all ran correctly but I did get a message saying that "training was complete for 5 iterations".
2/10: I was not actually able to find information in the proposal or the report on GCC for the 2016 Spring Tools Group. I did find out that the tools group last year installed Sphinx Trainer and Sphinx 3 although I did not see much information that went into detail in their proposal and report on the installations and details of the software. I also did not see really anything on GCC. I'm not sure if I am not looking in the right areas but I did not see anything in the proposal or report on that.
2/13: I noticed through reading this weeks logs that both Sharayah and Huong have done some research into G++. It looks like they have both figured out some good information on what the compiler can accomplish and some of the features. I also read that Huong noticed that GCC was installed on majestix from last semester. I did notice that we still need to do some research into Pocket Sphinx. Although Professor Jonas said he wanted G++ to take priority, I would still like to research Pocket Sphinx as well.
2/14: I have done some research online and looked at some links to G++ and GCC. GCC and G++ are the GNU compilers for C and C++ programs. It seems as though there are some misunderstanding of both GCC and G++ and them being two independent compilers. It seems as though GCC can compile both C or C++ programs and G++ can only compile C++ programs. The compilers differentiate program languages by checking the file name extensions. There seems to be a lot of information on installing G++ onto different systems. There seems to be a bunch of different websites that offer the download of the G++ file and libraries associated with it. I guess I will need to talk to my team and look at what they have found out and figure out which websites we want to download the files from and who is going to start with the installation process.
2/8: I want to make sure I take notes and write down more detailed descriptions on how to set up experiments and how to run a train. That was I can make copies and give information to the other members of my team so that they will not have any problems either if they decide to run sub experiments or run a train on their computers as well. I also want to talk to the group and discuss adding Torque to the system and see what the other member think.
2/10: Continue to research Pocket Sphinx. Continue creating a list of why Pocket Sphinx would be better to have installed on the system instead of Sphinx 3. Also do some research on G++. Professor Jonas said it would be cool to see if our group could install G++ onto a drone and then possible on to caesar if it would better the project.
2/13: Try to find additional information on G++ for the coming week so that I can further research that has been done by both Sharayah and Huong. Both of them were able to find a little bit of information. I would like to be able to further the research of what they have found so that we can add it to our final proposal and also have enough research for why it would benefit us to implement it. I also would like to research more on Pocket Sphinx to further explain why we should upgrade to Pocket Sphinx when Sphinx 3 is currently installed.
2/14: Talk to the systems group in class on Wednesday 2/15/2017 and determine if the 2016 Tools group captured their installation process of GCC onto one of the drones. If they did capture it then it would benefit our team to have that information. If not, we will have to reinstall GCC or install G++ on our own and capture the installation process and document our findings.
2/8: I am a little concerned that the train did not run fully for me and that I did not do all of the steps correctly. I am interested to see what other people received for messages that ran a train over the past week so that I can compare and make sure it completely ran. Some of the classmates were saying that the train they ran took a couple hours. When I ran the train, it only took me about an hour, which makes me believe it did not fully run and I may have ran into an issue.
2/10: Not too many concerns at this point. I was able to find some information on G++. I am starting to get a little understanding of the compiler however I am still not sure everything it entails and how difficult it would be to set up on a drone. I also did not find out yet if G++ is backwards compatible with GCC so I will have to continue doing some research on both compilers.
2/13: I do not really have any concerns at this point. I am finding some information on both Pocket Sphinx and G++ that I think would help out with installation in the future. I do get nervous when installing new software because I want to make sure I am installing it correctly and not ruining anything. However, I think our group has been doing a lot of work reading on both Pocket Sphinx and G++ to help make the installation process go smoothly and help make every team member comfortable with the installation process.
2/14: No new concerns. I do want to talk to other groups as well as Professor Jonas to possibly further investigate installing G++ and the process since GCC is already installed on majestix. It did seem from talking to Professor Jonas that GCC was installed however I am having a hard time finding the exact documentation. This may cause our group to need an untouched drone to reinstall either GCC or G++ so we can capture the installation process ourselves.
Week Ending February 21, 2017
2/16: Generate a report to give to Professor Jonas on why we should be installing G++ as a compiler. As a group, we have all found additional documentation on implementing G++ that makes it seem as though implementing G++ should not be too complicated on Majestix, since GCC is already on that machine.
2/17: Log on and read logs from the other members in my group as well as the other teams, such as Systems Group since we are now working with them to get a machine to us.
2/20: Work on finalizing our Project Proposal, as well as continue to work on the report that will hopefully get approved so that we can implement G++ onto Majestix and then hopefully if all goes well, implement GCC and G++ onto Caesar. I also need to go through Majestix and make sure all the files that we need are currently still on the system. Making sure all the files are on Majestix that our group will need will enable us to continue with the installation of G++. If not all the files are on the system, we will have to work with the systems group.
2/21: Log in and read weekly logs from other members of my team as well as from other classmates. I already have a good idea of how my teammates are doing, based on constant communication and reading their weekly logs. However, I would like to further investigate the work that the other teams are currently doing as well as the individual members of those groups.
2/16: Currently, our report is in progress. We are working as a team to write the report. I know we are going to need to implement G++ because that is our first task. However, due to the fact that Majestix is not up and running right now, we have created a report that details why we should be using G++ as a compiler. We also found some documentation, that will be published in our report, that explains how to implement G++ onto Mejestix. Because the 2016 Tools Group has already implemented GCC, implementing G++ should only be a couple new commands and should only add a few extra directories.
2/17: Was able to find out through Andrew's logs as well as Sharayah's logs that Majestix is now up and running. It looks like Sharayah was going to try and copy some files onto Obelix in the mean time but now that Majestix is up and running, our group should be good to go and start moving forward with some of our installations after we get the approval.
2/20: The report for G++ has been improved. There is not much information in the report but hopefully the fact that the installation does not hit too many files, and does not alter much from what is currently on the system from the 2016 Spring Tools Group installing GCC onto Majestix, will allow us to go and install G++. Also it does seem that the files that we expected to be on Majestix are currently on the system. I looked at what the Spring 2016 Tools group said was hit/affected by installing GCC and I believe all the files we need are there and we should be able to continue the process of installing G++ based off what last years group had done with the install of GCC.
2/21: It looks as though my group members are all doing well and are keeping up with their logs and are also continuously researching. That gives my some comfort because in the next week we are hopefully going to be getting the approval to go ahead and install G++ onto Majestix and then Caesar. It looks as though all the systems are up and running now and the other groups have started to get the ball rolling on their tasks. It should be interesting now to see all the groups continue to make progress and have the groups work with each other to complete their tasks.
2/16: Hopefully our report gets approved by Professor Jonas, as well as the rest of the class. We are also going to hope that Mejestix is up and running by next class. If that is the case, we should be able to implement G++ onto Majestix. If that goes well, we can create a report on the files that get hit by implementing G++. If all goes well, we will be able to implement G++ onto Caesar with approval.
2/17: Continue writing our report on why we should be installing G++ onto Mejestix as well as Caesar. Also continue working on our project proposal and make sure it is in unity with the other members of our group as well as every other team and people working on the project. Also I need to access Mejestix to make sure the system has all the files that we are expecting it to have from the previous semesters. If not we can talk to the Systems group to see if anything has changed.
2/20: Talk to my other group members about double checking the files that are currently on Majestix. If they can double check to make sure they believe everything we need is on Majestix, it will help confirm what I had looked over. Also reach out to other classmates who have not fully completed the Project Proposal. Hopefully they have the information they need to finalize their sections so everyone can help make it uniform and complete.
2/21: We are hoping that our report on the caparison of G++/GCC will get the approval by Professor Jonas this week so that we can install G++ on Majestix. The goal of installing it on Majestix will be to capture everything that the install process does. That way we will have a good idea of the install process of both GCC and G++ and then will be able to successfully install both pieces of software onto Caesar if we are told to do so.
2/16: I do not have many concerns at this time. The only concerns I have are that hopefully the implementation on G++ goes as smoothly as research suggests it should. I also have the concern that Majestix may not be up and running by next class. If it is not, we are going to need to copy the files on Majestix over to another drone so that we will be able to implement G++ onto a drone. I'm not sure how that process wil go with Sharayah, but hopefully all goes well with copying the files.
2/17: No concerns now that Majestix is up and running. This should allow us to get our report on G++ approved by next week and hopefully start the install of G++ onto Majestix. The only concern is that the install may be a little more complicated than we expect it to be. However, we have all been doing a lot of research on the install process and I feel as though our entire group is comfortable working on the install and we should all be able to help each other out if we run into any complications with it.
2/20: I do not have too many concerns right now. One of my concerns though is that not everyone will finalize their portions of the project proposal. The proposal is due in a couple days and the proposal needs to be read by everyone and made sure that it is uniform and complete so that it flows nicely and contains all of the information that each group intends on contributing to the proposal.
2/21: I do not have any concerns right now. I am actually feeling pretty confident. I am hoping that our group will be able to go ahead and install G++ on Majestix and with all the research that our group has been doing on the software and the install process, I feel as though we should be able to get through it smoothly. Even if we have a little problem, I think we have all been trying to research as much as possible so that we will all be able to work through it together and get the install complete.
Week Ending February 28, 2017
2/23: Try to go back into the experiment and successfully run train and decode. I was having trouble before running the train and decode, but had recently talked to some classmates and feel as though I have learned a little more about the train and decode and will have a better understanding of the direction that were a bit unclear to me before. Successfully running the train and decode will then give me a better understanding of how the system is working and will allow me to further my knowledge of our system as well as benefit myself in the future when running more experiments.
2/24: Log into Majestix, unmount Majestix, and attempt to complete an installation of G++ onto Majestix. Trying this early in the week will give me a little wiggle room in case I am having problems, which I am sure I will, because I usually do not find that things work perfectly on the first attempt.
2/27: Log in and read logs from my group teammates as well as other classmates to see how progress is being made throughout the week and to see if I can be of assistance to anyone who may be struggling to figure anything out.
2/28: Log in and read weekly logs of my classmates as well the weekly logs from my group members. I am hoping that I can gather information as to what they are currently working on, what problems they currently have, and if any of them could be of assistance to me in the struggles that I am currently having right now with my tasks.
2/23: I was successfully able to fun a train and decode and have put my results in a table under in my sub-experiment folder https://foss.unh.edu/projects/index.php/Speech:Exps_0299_001. I am very happy that I was able to complete the train and decode successfully. I learned a lot from doing this and also feel more confident for future experiments that I will have to run individually as well as in groups. Having a better understanding of the system and how the train and decode works will also allow me to further help my classmates as we continue to advance the system.
2/24: So I did end up running into problems when attempting to get onto Majestix. The first thing I tried to do was to ssh into Majestix. I was able to do that with no problem. However, the second part is where I am currently running into problems. I tried to perform a "umount -a" command on Majestix to unmount it from Caesar. I am currently getting the error message stating that "umount: only root can do that". I have never got this error message before. Earlier when I was trying to perform the "umount -a" command, I was receiving an error stating that the device was busy. Now I am receiving an error message stating that I can't even perform the command. I figured that I could possibly perform a sudo command to unmount that device. However, when I tried to perform the sudo command, it prompts me for the sudo password for my accound "[sudo] password for jaa334". I attemped to type in both my password and the root password but neither of the passwords are working. I am kind of stuck at this point because I know for a fact we need to unmount the machine from Caesar before trying to perform any installations so we do not mess anything up on Caesar.
2/27: It looks as though Huong was able to copy the files over from Caesar to Majestix. It also looks like my group members, as well as the rest of the class, have all been running trains this week to familiarize themselves with the system. I also noticed that Majestix still does not seem to have any internet access.
2/28: It does not look like any of the classmates can be of assistance to me at this point. It is a little frustrating because I do want to get work done and make sure I continue moving forward on my tasks. I have been doing a lot of reading on why I am running into the problems when trying to unmount Majestix right now. I may have to talk to Professor Jonas in class if I cannot figure it out on my own of with the help of any of my classmates.
2/23: Now that I have successfully run a train and decode, my next task will be to work with Huong and try to install G++ onto Majestix. I know that we were having some problems yesterday unmounting the machine, however I am going to do some research into why we were getting error messages stating that the "system was busy" even though we were the only ones on the machine. Hopefully I will be able to figure this problem out and will then be able to continue going forward and attempt to install G++ onto our drone. If this works out well and the install goes as planned, we will be able to capture the installation and use our notes if the class decides it will be beneficial to install the compiler onto Caesar.
2/24: I am going to have to try and talk to my group, as well as do some research, to possible figure out why I am getting this error message and to figure out if there is another way to untie Majestix from Caesar. I definitely want to continue trying to install G++ onto Majestix but I do not want to do that until I figure out how to untie the drone from Caesar.
2/27: Continue with the installation process of G++ on Majestix. There are a couple things standing in my way as of right now. Once being that I am having trouble with unmounting the drone from Caesar still. I am not sure why considering Huong and Professor Jonas were able to do it the other day in class. I am getting a different error message though which is not allowing me to unmount it. The other interference is still that Majestix has no internet access, whereas Caesar does. I will continue to research how to possibly install G++ without internet access, due to the fact that all of the research we had done and the documents we found implied we had access.
2/28: Continue working on the installation process of G++ on Majestix. I have been running into problems and have not been able to resolve them over the past couple days. I am hoping that I will be able to work through the problem. If not, I will have to talk to Professor Jonas and figure out if he has any insights to my problems. I also would like to help assist Sharayah if possible because I know she is currently trying to run a train on Majestix but is failing in the decode steps.
2/23: I do have a few concerns going forward with this week. I am going to attempt to install the G++ compiler onto Majestix, however, we have already been running into problems with the machine before we can even do the install. We were having trouble unmounting the machine from Caesar and we were also told that there was no internet access going to Majestix. We were told that we could not install anything onto the drones unless the drone had been unmounted from Caesar and we our group was also under the impression that we would need internet access during the installation process. I am hoping that these wont be an issue and that we can install the compiler because that is a task that has been put at the top of our list and I do not want to have to put the task off any longer.
2/24: I do have multiple concerns now at this point. One of my concerns is the fact that I cannot figure out how to untie the Majestix drone from Caesar. I thought that this process would be relatively easy but it just seems to continue to be more complicated every day I try it. I am going to have to continue doing research to figure this out. After I am able to untie the drone, I do have concerns that I will not be able to figure out how to install G++. I assumed that Majestix had all the files we would need to continue the installation of G++, however, it looks like I will have to download some tar files and put them onto Majestix to be able to install G++. I was also hearing from my group members that Majestix doesn't have internet access so I am concerned as to how I will be able to access those files and put them onto Majestix.
2/27: The same concerns now as I had on 2/24. I am having trouble with the installation process, receiving errors stating that I cannot unmount Majestix. We also are without internet access on our drone. The research and documents we found implied we had access to get the files needed for the download and install the compiler on the drone. I am hoping that my other group members will be able to help me out or we may possibly need to meet up and figure out a solution to installing G++ on Majestix.
2/28: Same concerns now as I've had most of the week. I am putting a lot of time and effort into figuring out why I cannot unmount Majestix. I even tried doing so as root but when I look into the folders, I still can see files listed in them which makes me believe I have not successfully unmounted Majestix from Caesar. I will continue to work with my classmates, and Professor Jonas if possible, so that I can resolve this problem and continue moving forward with my tasks.
Week Ending March 7, 2017
3/2: Run and hopefully succeed on running a train and decode on Majestix. This is important for our team to run the train and decode on Majestix because our plan is to be able to install software tools on Majestix. However, we want to make sure that we have an initial train and decode successfully run on Majestix. This will allow us to install software, and then run another train and decode, and then compare our two results to make sure that the software we installed did not mess or change any of the numbers that should be returned after a train or decode on Majestix.
3/3: Log in and read weekly logs of other classmates as well as those in my group. We are supposed to copy files from Majestix over to Caesar, incase the systems group decides to completely wipe clean Majestix. I can read the logs of my group members to see if anyone has located files as well as if they have copied those files over to a safe location on Caesar.
3:6 Checking in and reading logs of my classmates and other group members. I would like to see where my group members are at and if they have done anything with the files on Majestix. If they have not, I will make that my task for tomorrow.
3/7: Copy over files from Majestix to my home folder on Caesar so that the files can be backed up and saved if anything needs to happen to Majestix, such as wiping it clean and rebuilding it. Because we do not have packages that we need on Majestix to move forward with installations, there is a possibility that our group is going to be assigned a new client in the upcoming week. If this is the case, we need to have all the files from Majestix so that we can use them going forward, as well as have the information to compare installations.
3/2: Originally our team was able to run the train on Majestix but not the decode. We were having trouble and it took the entire team and the help of Professor Jonas to realize that when we copied over files from Caesar to Majestix, we had copied over too many files. This actually made us run into errors when trying to run the decode because the decode could not find the correct file and function to run. Therefore, we decided to run the decode on Caesar with the train data that we had run on Majestix. We know this was not the correct way to do it but we decided to work around our problem. We did end up getting the problem fixed and deleting some of the files that we did not need. This ended up allowing us to be able to run the decode on Majestix now. Our results can be found in the experiments under /mnt/main/Exp/0299/007 folder or online at ~> https://foss.unh.edu/projects/index.php/Speech:Exps_0299_007
3/3: After looking at Sharayah's logs as well as the logs from the 2016 Tools Group, it appears as though some of the files that should be on Majestix are not on Majestix. These files include one's that were added with the installation of GCC onto Majestix.
3/6: It does not look like my group members have copied any files over to Caesar. That is okay though, I can copy them over to Caesar by Wednesday's class so that if the system group decides to wipe Majestix clean and rebuild it, our group will have all the information we need from past semesters to relate to and to use for our work this semester and once we get a new client.
3/7: I am not really sure what happened with the past semesters but it does not look like their are any files on Majestix that we need to copy over to a group member's home folder on Caesar. I logged into Caesar and then ssh'ed into Majestix. I went through all the directories and could not find any files that I would need to copy over to Caesar. I did try to change directories into a user's home directory from last spring, however, my permission was denied. I was thinking that it could be possible that one of the students from last year may have saved files into their own home folder on Majestix. However, since I do not have permission to go into their folders, I decided to try to access their folders as root. My permissions were still denied as I tried this. I doubt that they would have saved anything under their individual folders, if it was useful and public information about installs and Majestix. Therefore, I am assuming that the files that we would possibly need may have been erased or maybe the previous semester did something else with the files instead of keeping them on Majestix.
3/2: We actually just found out that we may need to be switching over to a new client. Also there was some miscommunication between our group and other people. Origonally we just figured out what files had been hit from installations on Majestix before in past semesters. We misunderstood and now we need to go in and figure out what textfiles are on Majestix and we need to copy over those textfiles to a local place under our home pages so that we can possibly save those text files if Majestix needs to be cleared completely.
3/3: Continue to dive into Majestix and find files that are important and that we need to keep from past semesters. If we can locate these files, we need to protect them incase of a complete wipe of Majestix. Therefore, Professor Jonas has asked our group to copy those certain files over to a safe place on Majestix under one of our individual home folders.
3/6: copy over and back up any files that are on Majestix that are important. This will give us the ability to have the files if the systems group decides to wipe Majestix clean. Then we can put these files onto the new client, if we get one, and continue using them as if they were on Majestix. It is important that we do not lose any information from Majestix if anything happens because our group will need those files that are currently on the Majestix client.
3/7: I found out from a couple other group members that they could not find any important files that we may need on Majestix either. Because we were told we may need to switch clients and use Obelix, it would be a good idea for everyone to run a train and decode on Obelix. It looks as though Sharayah and Huong have already done so. I will also run the train and decode so that we can all make sure that we are on the same page and we know how the train and decode runs on Obelix.
3/2: I do have actually a big concern at this point. So it looks as though Majestix may be broken/there may be problems with it that we are running into that could cause us to not be able to complete our work. It looks as though the files and packages that should be on Majestix are not currently on the client. Therefore, there has been talk that we may be switching over to a new client. If this is the case then we are going to have to put in a lot of work and effort to catch up and redo some of the things we have done with Majestix on whatever new client we get. If we cannot get a new client, we are going to have to figure out why Majestix is different than the other clients and what we can do to fix it.
3/3: Same concerns as of 3/2. It is important that we find those files and back them onto Caesar. Also it is a concern for our team that we may be dealing with a new client if the system group decides to wipe Majestix clean. That could mean we may need to install GCC and G++ on the new client as well as locate important files on the new client and run a train and decode on the new client to make sure no results are changing when we just do an install of a new software.
3/6: No new concerns today. I basically have the same concerns as of 3/3. Hopefully we will discuss in class this coming up week what we are going to do with Majestix client and go from there. That should give our group some direction and further information.
3/7: I do have a couple concerns. One of my concerns is just switching to a new client. I am not sure if we are actually going to be switching to a new client but it would be nice if we could, unless Majestix has been fixed (which does not seem to be the case), just so we can go forward with a couple of our installations on the clients and then install software onto Caesar. I also have concerns as to what the spring 2016 tools group may have done with their documents from their semester and their installations on Majestix. Three of our group members have gone through Majestix and none of us can find any files. I am not sure if the files have been erased for some reason or if the group last year put them somewhere else and did not document it. It would be nice to have those files to use going forward for our records are we begin to install software on a client or caesar.
Week Ending March 21, 2017
3/8: Now that our group has been assigned a new client, Obelix, and we were able to run a baseline train and decode on Obelix, we decided that we were going to go ahead with the installation of GCC on Obelix. Because we need to install GCC before we can install G++, we are going to complete the installation of GCC on Obelix. If all goes well, we will create a documentation report, as well as run and train and decode on Obelix after the installation process to make sure that the results from the train and decode have not been affected from the installation.
3/9: Now that GCC has successfully been installed, I am going to take everything that we did in order to install GCC on Obelix and put that into a documentation page so that other classmates can see how we installed the software on Obelix. Also, for future semester, those students can see what we did to install GCC on Obelix and the results that we got from doing the installation our way.
3/20: Log in and read logs from my Tools group members as well as from my new team members. Attempt to see if any members have continued the work involved in the documentation and running experiments after the implementation of GCC on Obelix.
3/21: Log in and check logs from my group members. Figure out if they are having difficulty and/or need help completing any tasks that they were trying to accomplish over the past week.
3/8: The installation process went quite well actually and Obelix now has GCC installed. We did run into a little complication at first with the installation process. When trying to install GCC, we were getting an output saying that there was an "Error: nothing to do". We then thought that we may possibly need to first update, however, we were still getting an error stating "Error: no packages marked for update". After doing some research, we found that we may need to check the time and date that the client had set. The date and time were off by over 9 years. We then reset the date and time, tried running the installation commands again, and we were successfully able to install GCC on Obelix. Our documentation and results can be found at ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_GCC_Install_Documentation
3/9: I have updated the blank page I created yesterday ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_GCC_Install_Documentation. This page now contains the information and steps we took in order to install GCC onto Obelix. The documentation also shows the results of the packages that were updated as well as installed. I have also added the errors that occurred in trying to install GCC and how we overcame those obstacles in order to successfully install GCC on Obelix.
3/20: It looks as though my group members are satisfied with the documentation that went into the installation process of GCC on Obelix. It also looks as though Sharayah intends to run a train and decode on Obelix. This will be helpful to see if any of the results have changed after the isntallation of GCC on Obelix. Hopefully the results are not different than the baseline experiment which gives us clarification that the install of GCC did not disrupt anything.
3/21: It still looks as though only Sharayah has attempted to run a train and decode on Obelix after the installation of GCC, according to her weekly logs and the logs of my other team members. It is important that all of us try to run an experiment so hopefully we will all find time to do so before our next class so that Sharayah is not the only one who attempted to run an experiment for evaluation.
3/8: Now that we have successfully installed GCC onto Obelix, we are going to want to run a train and decode. We will then be able to compare the results to our baseline train and decode to make sure that the results from those were not altered by just installing new software. We will also want to create a documentation page so that other people and future semesters can see how the installation process went, as well as the packages that the installation process created or upgraded.
3/9: The plan now is to take snapshots of the file system in Obelix and put that into a text file so we can compare that with the snapshot of the file system that we had on Obelix before we installed GCC. We are also going to want to try and go forward now with installing G++ onto Obelix. After that, we should be able to get approved or denied from Professor Jonas on whether he wants us to go ahead and install that software onto Caesar.
3/20: Continue checking in and working with group members and team members. Also try to run a train and decode on Obelix myself to match it with the baseline experiment that we did before installing GCC to make sure that the results are still consistent. Also go further ahead and work on the installation process of G++ on Obelix.
3/21: Because Sharayah was able to successfully run and train and decode on Obelix, on the seen data, it may be a good idea for us to run a train and decode on unseen data. I am not sure how many of us have run the train and decode on unseen data so it is important we do so, that way we can have more information on the baseline experiment results for Obelix that we can use to compare after each new software installation.
3/8: I actually do not have any major concerns at this time right now. Our installation process of GCC on Obelix went very well. I am certain we will be able to run the train and decode on Obelix to make sure the results are not different from our baseline train and decode on Obelix. That will further reiterate that just installing the software did not change the results that we should be getting from running and train and decode.
3/9: Still no major concerns right now. The install and the documentation went well for installing GCC onto Obelix. The only concern I have is trying to install G++. We would just need to make sure that we have internet access. We can double check with the systems group before trying to install G++ to make sure we do in fact have internet access still. If we do, we can go ahead and try to work on the G++ install and documentation on Obelix.
3/20: No new concerns today. It looks like my group members are happy with the installation process and documentation of GCC on Obelix. Hopefully myself and the other members of our group will have time to all run experiments on Obelix so we can all compare our results against our initial baseline experiment on Obelix before we installed GCC.
3/21: No new concerns as of right now. It appears we were able to get a lot of the work done that our group wanted to get done before our next class. The only concerns I have are the tasks that our new team is supposed to be doing. Because only half our team was in class last time, it made it difficult for us to figure out the direction we wanted to take from here on out as a team. Hopefully we can all talk together in our next class and really establish a direction for the rest of the semester and our project.
Week Ending March 28, 2017
3/23: Install G++ onto Obelix and document everything that occurs for future reference when installing the software on other machines, as well for future semesters to know anything that we encountered during the installation process.
3/24: Create the installation documentation for G++ on the wiki page. This will show classmates how we installed G++, problems we may have run into, as well as solutions for solving those problems so that other people can use this document as a reference in future years to come.
3/27: Log in and check the logs of other group members and other team members. This will give me an idea of where my group stands for their tasks for the week, what progress they've made, and what issues they are having. Also I hope to gain some knowledge as to if our team was able to run a successful train and decode and to figure out if our team's client is working the way we expected it to.
3/28: Checking in and reading logs from my group members and my team members. My group seemed to be doing well on their tasks of of last time I checked in so I am not too worried about them. My team members also seemed to be doing well but I will see if there is anything I can assist with or even tasks I can pick up on from reading other entries and logs.
3/23: The installation process was able to be successful and I was able to install G++ onto Obelix. We did run into a couple issues. One of the issues being that last week we thought we had already installed GCC onto Obelix. However, it looked as though we actually did not successfully finish the installation of GCC on Obelix because we were not able to find it anywhere on the system. We had to actually go back and do a yum install of GCC onto Obelix and then checked the version we were using of GCC to make sure it was installed. After that we went forward with the installation of G++ and we were then successful and able to finish the installation of G++. We made sure we had it installed correctly and were able to find the version we were using on the client.
3/24: The G++ documentation page has been created on our wiki site at ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_G%2B%2B_Install_Documentation. The documentation page goes over the steps that we used to installing both GCC and G++, as well as checking versions, making sure we had successfully installed the software, and the packages that were either installed or updated to the system. The documentation is not quite done though because we still need to add the list of files that were hit by the installation process and take a snapshot of those files, as well as add that documentation and the steps we did to complete that to the G++ install documentation page.
3/27: It looks as though Sharayah has taken the time to run the train and decode on Obelix to check to see if the results of the decode are any different than the baseline results, since we have recently installed G++ on Obelix. Also it looks as though Vitali has been able to run a train and decode on Asterix successfully. This confirms our notion that our client is up and running well and we will now be able to go ahead and implement a plan as to how we want to further develop a world class speech recognition system.
3/28: It looks as though our group is managing all of our tasks well. Although I did not see it in the logs, I did have communication with Jeremy and he did say that he took a snapshot of the files in Obelix after the G++ installation. The snapshot has not been documented yet on the wiki page so it will be important to do that. Also from reading logs of my team members, it looks as though other classmates are running into trouble running a 5hr train on Asterix. They were all able to run the 30hr traina and decode, but the 5hr seems to be failing during the decode.
3/23: The plan from here would be to put our notes of the installation process into documentation and add it to the wiki page. It is important to create documentation so that every other classmate can see exactly what we did, see problems we may have run into and how we solved them, as well as use it for reference if anyone else has to install the same software.
3/24: The plan for here to is to take a snapshot of the files that were hit during the installation process of G++ on Obelix. This will allow us to see what files have been changed or added to the system that may not have been there on Obelix before we installed the software. We also are going to want to perform another train and decode on Obelix to make sure that the results are still consistent with the baseline results we received a couple weeks ago. Because we have not changed the method of trains and decodes yet, we should not be receiving any different results from the baseline after installing a couple different software tools.
3/27: Talk to my team members to see if they need anything from me or need my help with anything. Also it is important that we take a snapshot of the files that were hit by the G++ installation on Obelix. I thought Jeremy said he was going to take care of that and add the documentation and results in our G++ documentation page, so I will check with him on that and hopefully he was successfully able to accomplish that task.
3/28: Talk to my group members and get the step by step instructions and results of them taking a snapshot of the file system in Obelix. I will then translate that onto the wiki page and into our G++ Installation page so that we have it documented as to what files were hit and what files may have been added.
3/23: I do not really have any concerns at this point. The installation process of G++ went well, I think that the documentation portion will be done within the next couple of days, and that we will be able to run train and decodes to check the results after the installations and to make sure that they are still consistent with the results we have from our baseline experiments.
3/24: I do not have any major concerns at this time. I guess one concern would be that I am hoping none of the results from the train and decode that we need to run on Obelix would be different from the baseline results that we took a couple weeks ago. If the results are significantly different than the baseline results, then we know that installing GCC and G++ may have caused something to be different in the train and decode results that we do not want and need to take into consideration.
3/27: No new concerns now. It is going to be important that this week we come up with a strategy and plan to further develop our speech recognition system and to try and figure out a way to reducing the word error rate. Hopefully we can discuss a strategy and start implementing that strategy this coming up week.
3/28: No real concerns as of today. After seeing some of the logs though, I am wondering why some people are running into trouble running a 5hr train and decode on Asterix. It seems as though it is failing during the decode. This is not a problem though when running a 30hr train and decode. Maybe I can run an experiment myself and figure out where the issue is occurring and then figure out what may be causing the issue.
Week Ending April 4, 2017
3/30: The Modeling Group had asked us if we could install miniconda onto their two client machines. Miniconda is a software tool that contains conda. Conda is a cross-platform package and environment manager that installs, runs, and updates packages and their dependencies. It allows the user to easily set up and switch between environments on the user's local computer. Miniconda is the same as anaconda but is stripped of everything except the python and conda package installer. Installing miniconda on the clients will allow us to further install numpy and scipy on those clients as well. The Modeling group asked us to install numpy and scipy so that they can perform their LDA tasks on those clients.
4/1: Logging in to check weekly logs. Try to see if I can find out if Huong or someone from the Modeling Group was able to fix the problem of accessing python through miniconda. If nobody has fixed the problem, it will be a responsibility to fix the problem of accessing python through miniconda since I was with Huong when doing the miniconda installation on Asterix.
4/3: Checking in and reading logs from both my group members and my team members. I have not had time to complete any additional tasks for the week since I have last checked in. Therefore, I am just going to read logs and figure out what my group members have done, if they've installed any new software, and if they are having any troubles.
4/4: Create the documentation page for the installation process of Miniconda on Asterix. This will allow other classmates to see what steps we took to install the software, what problems we ran into, what problems we are currently involved in, as well as for other classmates to offer direction to solve some of the problems that we are currently having.
3/30: We were able to install miniconda on Asterix. We followed the tutorial on the website ~> https://www.atlantic.net/community/howto/install-python-2-7-centos-anaconda/ and just made sure that we installed the 32 bit and not the 64 bit version. Everything went fairly smoothly and we did not run into too many problems. We just had some basic user errors that caused us to run into some trouble, such as running the 'wget' command on 'https://repo.continuum.io/miniconda/Miniconda-latest-Linux-x86.sh' where we should have been running the command on 'https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86.sh'. We went directly to the website to figure out there had been a slight change in the url, but after figuring that out it was fairly straight forward.
4/1: So it does not appear that anybody has fixed the problem of accessing python through miniconda on Asterix. The Modeling Group asked us to install miniconda and numpy and scipy so that they could use the software to perform their tasks regarding LDA. It also looks like nobody has taken a snapshot of the files either of Asterix to see what files were hit.
4/3: From reading the logs, it looks as though we have not installed any new software or fixed any past installations that we were having trouble with. However, it does appear that we do now have a list of what files were affected by the G++ installation on Obelix, which can be found here ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_G%2B%2B_Install_Documentation. Also, it appears that our group has started putting together a document and a proposal for why it may or may not be necessary to install Pocket Sphinx on a test drone and then on Caesar.
4/4: The results and steps from the installation of Miniconda on Asterix can be found at ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_Miniconda_Install_Documentation. The documentation is not completely done because we are still running into problems right now trying to install numpy and scipy. Once we get those problems figured out, we will then update the documentation page so that is is complete and continues to show each step we took to solve the problems and finish the installation process.
3/30: Now that we have installed miniconda on Asterix, we need to install miniconda on Miraculix for the Modeling Group as well. This should be fairly easy now that we have already completed the installation once on Asterix. We also need to take a snapshot still of the files that may have been affected by the miniconda installation on Asterix and make sure that we are not destroying anything on the client.
4/1: Most importantly we need to fix our installation problems regarding miniconda on Asterix. As the tools group, it is our responsibility to install software for other groups when needed. That also comes with the territory of fixing problems with software for those groups as well. Since we installed miniconda on Asterix, we need to be the ones who solve the problem so we can continue installing numpy and scipy. Also we should be taking a snapshot to see what files have been hit by the installation process as well.
4/3: The most important thing right now is to get scipy and numpy isntalled on Asterix for the Modeling Group. They do need these for LDA. Then after that is complete we can go ahead and attempt to install miniconda on Miraculix. We can also continue working on the documentation/proposal of Pocket Sphinx and whether or not we should attempt to install it.
4/4: We need to try and figure out a way to solve the path issue so that Asterix can access python 2.7 through miniconda. Once that issue is complete, we will install numpy and scipy on Asterix and update the documentation page. Then we will know exactly what steps to take and can then install miniconda on Miraculix. We can also continue to work on the Pocket Sphinx proposal to see if it would be worth our time and efforts in trying to install Pocket Sphinx on a drone.
3/30: Vitali had mentioned a concern about where we saved the installation on Asterix. We ended up saving the files in the root/usr/bin on Asterix and we tried to make sure that we did not overwrite any of the python files that were already installed on Asterix from the modeling group members. Vitali voiced some concern about the path of the file potentially causing a problem with python but he did not seem too sure and would talk to us if the path does bring up any conflicts in the future for the modeling group.
4/1: My main concern is if we have messed something up with the path on the Asterix client. Their drone was working perfectly fine before Huong and I tried installing miniconda, therefore it is our responsibility to fix the path so that they can access python through miniconda and are able to continue with their tasks regarding LDA.
4/3: Same concerns as last time. I am worried that we may have messed with the path on Asterix when installing miniconda. Therefore, we cannot install numpy or scipy on that drone. Hopefully we can get the path issue sorted out so that miniconda can access python 2.7 and our group can continue installing software that other groups/teams may need.
4/4: At this point I basically have the same concerns. Our big task that we will have to solve tomorrow is how to figure out the installation process for scipy and numpy after we sort out the problem we are currently having on Asterix. We will be able to talk with the modeling group as well, who works on Asterix, to see if they could possibly work with us in finding a solution for the problem we are both having. Then they would be able to continue working with LDA and miniconda to complete their future tasks.
Week Ending April 11, 2017
4/6: Attempt to install miniconda on Miraculix with Huong as well as work on the poster presentation for the Undergraduate Research Conference. Last week we were able to install miniconda on Asterix but we ran into some issues with that drone running python 2.7 after the installation process. For the installation of miniconda on Miraculix, we are going to be more careful with the path of the file so that we are not messing with anything on the drone and that the modeling group will still be able to access python 2.7 after the miniconda installation.
4/9: Checking in and reading logs of my group members to see if any of them have made any progress on the URC poster, the miniconda installation on Miraculix, or any other tasks. We had created a rough draft of the URC poster, so I should be able to read logs to see if there is any other work to complete on it. Also, if I read Huong's logs, I should be able to see if she has done more work on the miniconda installation and whether or not she was able to give any hints of why we could be running into errors while trying to install the software onto Miraculix.
4/10: Install miniconda on Miraculix. Huong and I were having a bit of trouble trying to install it last time that we attempted the installation processes. I am hoping that this time, the process will go smoother and I will be able to install the software correctly. That will enable the modeling group to continue working on Miraculix and use the drone as part of their tasks with LDA.
4/11: Huong copied the Miniconda file over to miraculix last night while she was at school so that we would not need to worry about having internet access to install miniconda. Therefore, my task will be to just go into the folder that the miniconda file is located, turn the file into an executable file, and then I should be able to go straight forward with the installation process.
4/6: In order to make sure that we did not mess anything up, Huong decided that it would be best if we just copied the executable file of miniconda over to Miraculix, unpackage it there, and then just run the file to install the software. We were having trouble with this due to the fact that the drone could not find the correct folder destination that we were specifying. We know that the folder was definitely there and existed but the command continued to give us error messages back. We are going to continue to look into the problem so that miniconda will definitely be installed on Miraculix in the next couple days. Also we were able to put together a good starting point/rough draft for our URC poster. We feel as thought the poster definitely contains the information necessary so that the audience can understand the project as well as what our tasks have been as the Tools Group.
4/9: After reading Huong's logs, it looks as though she was not able to figure out a solution to the error message we were both receiving while we were trying to complete the installation process. She was able to figure out while the file directory error message was being given to her but then continued to run into another error message while trying to run the executable, but it looks as though we just need to run the 64 bit version on Miraculix to complete the install, whereas we ran the 32 bit version of miniconda on Miraculix.
4/10: I attempted to install miniconda on miraculix but ran into an issue. Miraculix does not have internet access at this time so we cannot go out and get the package we need to complete the installation process. This error is not major because I already know I am getting an error when trying to install miniconda just because the drone has no internet access. I have reached out to the systems group and am hoping that miraculix will soon be able to gain internet access. After that, I will be able to go on with the installation process of miniconda on Miraculix.
4/11: The installation process for Miniconda on Miraculix went just fine and now the software is installed. I was able to change the file into an executable and was able to run it just fine. I tried to avoid the pathing problem that we had last time we installed miniconda on Asterix, so I made sure I specified the path as /root/usr/local/. I did not do any further installations after miniconda, just because I was not sure if the modeling group needed any further installations after that.
4/6: Our first priority is to figure out why we are not able to bring the miniconda executable file over to Miraculix to run it on that drone. I have dont reading online and it definitely seems that we are using the correct command and we have double checked multiple times to make sure the file destination exists. Once we figure this problem out we should easily be able to run the miniconda executable on miraculix to finish the isntallation process.
4/9:See if I can get any information from Huong on to why we need the 64 bit version instead of the 32 bit version of Miniconda. I should be able to take what she has already done, and just use the 64 bit version and finish the install process on Miraculix. I want to be able to have the install done in the next couple of days so that the modeling group can continue working on Miraculix with the correct software that they need.
4/10: Hopefully I will be able to get someone from systems group to give miraculix internet access. I know not everyone has the same schedule and it is not always easy to get each drone internet access when someone from a group needs internet access because not all of our schedules align perfectly. I am hoping that miraculix will have internet by tomorrow afternoon and that I will be able to continue with the installation process of miniconda then.
4/11: Talk to the modeling group and make sure that the installation process is not messing anything up with their group. Also see if they need any further installations based off of miniconda installed on miraculix. I also need to make sure I take another snapshot of the files in Miraculix to see if anything has been changed with the installation of miniconda. Also another experiment should be run for training and decoding as well to make sure the results are the same on miraculix as they were before the miniconda installation.
4/6: I do not have many major concerns at this point. The only concern I have is the error message Huong and I continued to get when we tried to install miniconda on Miraculix. We do feel as though we will be able to figure this out without a problem though and we should be just fine continuing with the install process so that the modeling group can continue working on LDA tasks on Miraculix.
4/9: I am not really sure why the 32 bit version of miniconda would not work on Miraculix. The 32 bit version seemed to work on Asterix, so it is a bit concerning that we could not figure out a way of running the 32 bit verison on Asterix. Although, as long as we get the software installed, I will be happy because that will allow the modeling group to continue using Miraculix and work on their LDA tasks.
4/10: I do not really have any major concerns. I guess the only concern I have right now would be making sure that miraculix can possibly gain internet access by tomorrow. I know I am giving the systems group short notice so hopefully someone will be able to give the drone internet. That way I will be able to continue on with the installation process. If not, I will have to make sure that I get miraculix internet access on Wednesday when we all meet for class and immediately go forward with the install process during our allotted class period.
4/11: I do not have any big concerns right now. The only thing I would be a little concerned about is having another pathing issue on miraculix, as we did on asterix, when we installed miniconda. However, I do not think that will be an issue this time since I made sure to specify the correct location to store the file. Hopefully talking to the modeling group in class tomorrow will help clear up any concerns and just give feedback on the installation of miniconda working on miraculix.
Week Ending April 18, 2017
4/13: Install NumPy and SciPy on Miraculix for the modeling group. Numpy is a fundamental package for scientific computing with python. It contains N-dimensional array objects, sophisticated functions, tools for integrating C++ code, and algebra and number capabilities. Numpy can also be used to seamlessly and speedily integrate a wide variety of databases. SciPy is a python based ecosystem of open source software used for mathematics, science, and engineering. SciPy is used with NumPy to build on the array objects.
4/14: Checking in and reading logs. Hopefully I can find information from my other team members if they have found any information on how many files that GCC installation hit. Also I am going to be reading my team members logs as well to see if I can gather any information on if they are making progress for our strategy to produce improved results on the train and decode.
4/17: Checking in and reading logs. Today, I am looking to see if any of my other group members have figured out documented how many files GCC hit when being installed on Obelix. This will be important for us to bring to Professor Jonas so he knows exactly what files the installation will hit if we install it on caesar. Also I am looking to see if anyone has added to our Pocket Sphinx proposal and whether or not it would be beneficial to install it this semester.
4/18: Gather more information regarding Pocket Sphinx. Gathering this information will help us as the Tools Group create a proposal whether or not we think it would be a good idea to install Pocket Sphinx. We have done some research before and we do know that pocket sphinx can be very useful, however, with the time remaining in the semester, we will have to weigh our options and if we think it is work installing with just a few weeks left.
4/13: The installation of NumPy and SciPy on Miraculix was fairly straight forward and simply. Both pieces of software were installed with no errors or problems. This should now allow the modeling group and the teams to work on their drones and continue working with LDA to better the percentage or word error rate.
4/14: As of right now it does not look like many people have done logs (which is fine since it is only Friday). However, this makes it a little difficult to gather information from reading logs since nobody has really done any logs that are giving me the information that I was looking for while reading their logs.
4/17: It does not look like any of my group members have documented how many files the GCC installation hit on Obelix. I do know we did take a snapshot for comparisons so I will have to continue with that task. Also it looks as though Sharayah has done more research and has gathered more information for us to add into our pocket sphinx proposal for this semester.
4/18: We have created a proposal here ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_Tools_Group. As a group, we looked at pocket sphinx and a research project conducted by a group at Carnegie Mellon University. It seemed as though Pocket Sphinx was pretty close to the results of their baseline tests. Also word error rate was a little higher when using pocket sphinx. As a group, we thought pocket sphinx could be beneficial for the future. However, since we only have a few weeks left in the semester, we thought that Sphinx 3 being fine tuned to reach the best possible word error rate for this semesters project was a better option.
4/13: One of the big tasks we have to do is to figure out how many files were hit by the installation of GCC. We already know that G++ hit four files and is beneficial to have. We now need to figure out how many files GCC hit during the installation process so we can get it approved to be put on caesar. Once it is approved, we will be able to put GCC and G++ on caesar, and then most likely all the drones if necessary.
4/14: Work on the proposal of Pocket Sphinx and figure out how many files the GCC install hit. I am assuming we are going to be getting ready to install GCC and G++ next week on caesar, and possibly all the other drones if Professor Jonas believes it would be beneficial to go in that direction. I also want to create a documentation page on the correct way to install miniconda so that the installation does not get saved into the wrong default directory (which is the reason we were having troubles before accessing python 2.7).
4/17: I definitely need to figure out a definitive answer for how many files GCC hit when being installed on Obelix. I believe Sharayah took a snapshot of the files so I'm sure we will have a number for next class. Also I want to continue researching pocket sphinx for the proposal. If the installation process is easy and can benefit this semester, it may be a good idea to propose installing it to Professor Jonas. However, more research will be needed in order to figure out if it will be worth the time and effort.
4/18: The plan from here will be to talk to Professor Jonas and make sure he is on board with installing GCC and G++ on caesar. If he does give us permission to install both on caesar, then we will have to make that our group's first priority to get both installed. We will have to make sure we are very careful with the installations as well so we do not screw anything up on caesar.
4/13: No major concerns at this point. However, we were worried that we may have taken the snapshot incorrectly of Obelix when installing GCC. If a mistake did happen during that, we may not be able to figure out exactly how many files GCC hit. Although we should be able to look back at the past years logs to figure out how many files GCC when the tools group installed it on one of the drones.
4/14: No major concerns right now. Still the same concerns as of last time. I am really not sure if we took the snapshot of the files in Obelix correctly before we did the GCC installation. If we did not do it correctly, then it may be a little tough to figure out how many files the installation process hit.
4/17: No new concerns today. Basically the same concerns I have had all week. Our group needs to get the information on the GCC installation. Once we have that, we should be able to move forward with the installation of GCC on caesar and possibly other drones. However, that will be up to Professor Jonas to decide after we talk to him.
4/18: I don't really have any concerns. We were able to find how many files the GCC installation hit, as well as also creating a proposal to the benefits of Pocket Sphinx over Sphinx 3, as well as if we believe it is wise to install it this semester. Also, if we do go ahead with GCC and G++ on caesar, I will obviously just be a little nervous when installing new software on caesar. There has been a lot of semesters work and students have put a lot of time into this project, it can just be nerve racking because we do want to isntall software, however we just do not want to mess anything up.
Week Ending April 25, 2017
4/21: Logging in and checking logs. I am going to be reading logs to see if any of my group members have gotten word back from Professor Jonas on whether it is okay to install GCC/G++ on Caesar. I was told yesterday by Sharayah that she was going to send him an email, however, I have not heard back so I am not sure if she did send one or if Professor Jonas has just not responded yet.
4/22: Logging in and checking logs. Mainly logging in today to see if Professor Jonas has responded to Sharayah's email about installing GCC/G++. Hopefully he will have responded and has given us the okay and permission to install on Caesar. If he has, we will need to start as soon as possible on both of those installs.
4/24: Since our group has not heard back from Professor Jonas yet about the GCC/G++ installation on caesar, I am going to continue researching Pocket Sphinx, as well as Sphinx 4, since both could hypothetically be an upgrade from Sphinx 3, which is what is currently on the systems we are using in class this semester.
4/25: We still have not heard back from Professor Jonas in regards to allowing us to install GCC and G++ on Caesar. Therefore, I am going to work on my team tasks of creating a document of pros and cons for both LDA and RNN in order to help us narrow down the path we want to take when running our final 300 hour train and decode to get our final results.
4/21: It does not appear that many people have started their logs yet so I am not getting as much information as I would have liked by reading the logs today. I also have not heard any updates on installing software on Caesar yet, and since it is not in any logs, I have messaged my group to figure out where we stand in that process.
4/22: Professor Jonas did respond to Sharayah but she gave him a list of files that we hit and Professor Jonas wanted the snapshot of the files so he can go in himself and take a look at what was hit by the installations. Sharayah said she was going to send that out to him last night and hopefully he will take a look today or tomorrow and get back to us quickly so we can start with our installs that we need to get done for this week.
4/24: We have done some reasearch and have documented our research and sites as to why Pocket Sphinx could be useful to install on the system after this semester ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_Tools_Group. I was also able to find a site ~> http://cmusphinx.sourceforge.net/wiki/sphinx4:s3vss4performancecomparison that compared both Sphinx 3 and Sphinx 4. From what it looks like in this study, Sphinx 4, which is written in Java and not C (like Sphinx 3) look like Sphinx 4 could lead to a better word error rate by using sub-vector quanitization and limits the number of paths that cross word boundaries. Sphinx 4 also has some optimaziation that are not found in Sphinx 3 that include multi-threaded scorer, grow-skipping, acoustic lookahead, multi-level beams, and more tuning parameters. It does look as though this study found that Sphinx 3 was faster than Sphinx 4, although the trade in accuracy for speed with Sphinx 4 could be an upgrade for later semesters.
4/25: After talking to teammates in past weeks, looking at experiments, reading logs, and doing research, I have been able to put together notes and information that I believe could be beneficial to our team in preparing to run our final 300 hour decode. Hopefully our team will be able to have a final discussion tomorrow in class to finalize our plan for our last decode and how we plan to get the best results for the lowest word error rate to present to the class and to Professor Jonas.
4/21: Our main goal is to install GCC and G++ on Caesar this week. However, we cannot do any of that if we do not have approval by Professor Jonas. He had asked us to send him an email and I was told by Sharayah that she was going to do that yesterday. I have not heard back from her and I have not heard anything from Professor Jonas so I am not really sure if an email did get sent, or if Jonas has just not responded yet. We need his approval to complete our tasks so I am hopeful he gets back to us shortly.
4/22: The big thing we still have to do is get permission from Professor Jonas to install GCC and G++ on Caesar. Sharayah said she gave him the snapshots of what files were hit by the installs so hopefully we will be able to start installing those within the next couple days. If we don't have a response, we should continue looking at Pocket Sphinx and gathering as much information for our proposal so that following semesters can have as much information about Pocket Sphinx that would be helpful to them in making a decision in whether or not they should install it.
4/24: Hopefully hear back from Professor Jonas by tomorrow. If we do hear back by tomorrow, I will immediately go ahead and work on the installation process of GCC and G++. This will include taking snapshots of the files before and after each, as well as running experiments to make sure that we are not messing around with caesar and ruining the system. It will be very important for us to be as careful as possible so that nothing harms caesar.
4/25: Hopefully we will hear back from Professor Jonas in class tomorrow and can discuss any concerns he may have with us installing GCC and G++ on Caesar. I am hoping he gives us the okay to install both and the major tasks for next week will be to install both GCC and G++ on Caesar, as well as work on our teams plan on what path to take for our 300 hour experiment and our final results to present to the class.
4/21: My big concern is whether we are going to have enough time to install GCC and G++ on Caesar this week. I followed up with my group seeking information on whether we sent Jonas an email yet. If we haven't, I will take that into my own hands and email him as soon as possible. I need to hear back from Professor Jonas relatively soon so we can go ahead and install the software. We cannot be delaying this any longer because these tasks need to get done and they are our main tasks for the week.
4/22: My main concern is not having permission to install GCC and G++ on Caesar. That is basically our biggest goal as of right now and it would be a bad thing if we did not get permission. Hopefully Professor Jonas agrees that the installs do not him many files and that it is okay to install them on Caesar. I also have a concern that we did not take the snapshots correctly on the files before and after the installs and Professor Jonas is not going to be able to find the correct information he is looking for in order to make an educated decision on whether or not we can move forward.
4/24: Still the main concerns as I've had all week. Hopefully we hear back from Professor Jonas as soon as possible so that we can go ahead and install GCC and G++. I feel confident we will be able to install both software easily but will definitely be stressed and worried, since we are going to be installing software on the major server.
4/25: My main concern is not being able to install GCC and G++ on caesar. Installing that software has become our groups biggest and highest priority for the semester remaining and it would be a bummer if we effected Professor Jonas' decision to either install or not install the software in a bad way. We do think that it will benefit future semester so we would like to be able to present our thoughts to him as well as possible so that we are able to continue with that goal of installing the software.
Week Ending May 2, 2017
4/26: After talking with Professor Jonas, he has decided that due to time constraints that we are not going to end up installing GCC and G++ on Caesar this semester. His deciding factor was that he did not have enough information on what files were hit with the GCC install on Obelix. In order to make sure that GCC has not altered anything, he has asked us to run experiments on Caesar so we can compare it to experiments on other machines later.
4/27: Since Professor Jonas does not want us to install GCC and G++ on Caesar this semester, he stated that he was going to do the installations after the semester on his own. Because he is not an expert and we have been working on the project all semester, he asked us to clearly have a document that would explain both installation of GCC and G++ and each step we went through to install the software and what he can expect so it will be easy on him in the future when he goes to install the software.
5/1: Logging in and checking weekly logs of my group members and team members. I am not looking for much with my group members because we have been talking and have been working on our tasks. I am more so looking at my team members logs to see how our final experiment and report is coming along and to see if there is anything else I can do.
5/2: Logging in and checking logs. Basically just trying to communicate with team members today about running another possible experiment to see if we can better our results from our last experiment. I want to see if other team members have thought of any other ways we could reduce the word error rate and how we would go about doing so.
4/26: Our group decided to run a 5 hour train and decode as a baseline experiment for Professor Jonas to use in the future for baseline results before he goes and installs GCC himself. The results are located under experiment 0299/014 at ~> https://foss.unh.edu/projects/index.php/Speech:Exps_0299_014.
4/27: Our group has created a new document ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_GCC_G%2B%2B_Install_Documentation that combines both the GCC and G++ installation. Since you need to have GCC installed before you can install G++, we decided to combine the process into one page so that whoever is doing the install can understand how to clearly install GCC and then can move straight into installing G++ on the same machine without having to look anywhere else for instructions.
5/1: It looks as though we have run a train and decode on a large amount of data and have come up with a result that we are pretty pleased with. I am not sure if that is going to be the results we go with for our final team report or not. However, if that experiment has the best results we think we can get, we can go forward with the report and discuss our methods and results to give to the panel of teachers.
5/2: Our group has been in communication with each other over the past days and we may try and run another experiment on a large amount of data, but with a different method. I do not want to say the exact method on our logs because that could give it away to the other team but hopefully altering our method could give us a better word error rate.
4/26: Idefix is currently being used by a team to run a 300 hour train. When that train is done, we will run an experiment on Idefix to make sure the results on that baseline experiment are the same as the results we just received from our experiment on caesar. Also, since Professor Jonas will be working on the installation of GCC and G++ in the future, we are going to redo documentation pages to make sure that he installs the software in the correct order and is detailed enough for him to clearly understand the installation process so he will not have problems or questions.
4/27: The plan from here is to continue working as a group to figure out the best method and parameters to use in an experiment to achieve the best word error rate results. After we have figured a plan, we are going to want to run another train and decode, preferably on a large data set, and document our results and methods in the team's report.
5/1: Work on our team's report for our experiment. We have talked about the last experiment we ran and the results we received. Now we can put those methods and results in the report and further describe the process and why we chose to go with the method for running the experiment. We also want to make sure the report contains a lot of detail and explains clearly why we chose those methods to run the experiment.
5/2: Talk to team members and come up with one final method for running an experiment that we think could help reduce our word error rate. We did do a good job reducing the word error rate in our last experiment, however, we are not sure if the results are going to be good enough to better the Rebels. Hopefully altering our method and making slight changes could help improve the results a little bit and help us better the word error rate.
4/26: No major concerns right now. Since we are not going to be installing GCC and G++ on caesar, we are going to have to direct our focus to other tasks for the remaining of the two weeks left in class. I do not think that running more experiment on Idefix, working on documentation, or working on reports and our final team experiment should bring up any major obstacles.
4/27: Still no major concerns. The only slight concern I have is having the ability to run a train and decode on Idefix. The other team and modeling group has been running a 300 hour train on Idefix so that is taking up time on that machine and stalling us from using it. We wanted to run a train and decode on Idefix because Idefix has GCC installed and we want to measure the word error rate on that machine and compare it to the results on caesar to double check and make sure that the install of GCC does not affect the results in any way.
5/1: No major concerns right now. I am happy our team was able to run an experiment and we received results that we were pleased with. We can always choose to run another experiment to see if we can better our results, but even if we cannot better our results from this past experiment, we will be happy to report those results and explain our methodology for achieving those results.
5/2: I still have no concerns. The only slight concern I would have would be that our results from our experiment are not going to be as good as the Rebels results from their team experiment. However, we did use a method on a large amount of data that did produce results we are happy with. But since we have a week left, we want to try and see if we can alter the method a little bit to possible better our results and improve the word error rate.
Week Ending May 9, 2017
5/4: Work on documentation for the GCC and G++ installation process for when the software gets installed on Caesar. Because Professor Jonas will be installing the software later in the year, we want to make sure that he has the correct installation directions and everything is clear so that he does not mess anything up when trying to install both things on Caesar.
5/6: Checking in and reading logs and discussions from our team to see how the last experiment is coming along and if we are having any difficulties with running it with the different parameters that we specified. We decided to end up running a new experiment to see if we can better our results and I just want to make sure everything is going smoothly.
5/8: Logging in and checking logs of my group members and my team members. I reading logs as well as group discussions to try and figure out where my group is in the process of the documentation for GCC and G++, as well as to figure out where my team members are in our team report and possibly running another experiment.
5/9: Work on finalizing the final report for our teams final experiment. After talking with the team last week, we thought that the results that we got from our experiment may be a little bit worst that the other team. However, we were explained that the team with the best word error rate in their experiment is not always the competition winner. Therefore, it is important to explain everything about our experiment in detail, such as the methods we used, why we used that method for running an experiment, and the results that we received from that experiment.
5/4: The page for the documentation is located at ~> https://foss.unh.edu/projects/index.php/Speech:Spring_2017_GCC_G%2B%2B_Install_Documentation and gives clear directions for anyone who is not familiar with either piece of software and can still feel confident that they are doing so correctly. We also made sure that we added in taking snapshots so that the user can go back and check the files that were hit by the installation to see what has changed.
5/6: Unfortunately we have run into a problem while running our last experiment. We thought that improving the dictionary could help us reduce our word error rate. However, when we ran our latest experiment, we were having trouble with the new dictionary and some of the words that the experiment was picking up and cross referencing those words with the dictionary.
5/8: It looks as though Sharayah was able to run a 5hr experiment on Idefix. The results from this experiment are going to be used to compare the results to an experiment we ran on Caesar. The results should be similar and if they are similar, confirm that installing GCC and G++ on Idefix does not impact the experiment results and are the same as an experiment run on Caesar, which does not yet have GCC or G++. Also it looks as though Vitali may have had a mix up when running our final experiment for the Empire team and ran a 30 hour experiment instead of a 300 hour. Unfortunately at this point in the week, we are not going to have enough time to fun another experiment, so we will just have to use the results from our previous experiment.
5/9: The final report is basically done. We just have a couple other things to clean up in the report before we print it off and present our experiment and results to the panel in class. We believe that the report explains in detail our method in which we ran the experiment, why we decided to use that method and the results that were received from running the experiment.
5/4: The plan from here is to continue working on our team's final report. We, as a team, believe that we have run our last experiment, since we are limited on time now. We are happy with our results but we are not sure that the results are going to be good enough to beat team rebels. It is our job to make sure our report is well written and is very detailed with our methods and results so that we can convey how confident we are in our results and why we believe we ran a successful experiment.
5/6: I am not sure if we are going to have enough time now to run anther experiment, considering we were running an experiment on a fairly large amount of data. However, if we can tweak the dictionary to fix this error, it could help us improve our results and help us getting a better word error rate. If not, we should continue to work on our team report of our experiment so we convey our experiment well and have a lot of detail and descriptions to our methods and results.
5/8: Since we are not going to be able to run another experiment for the Empire team on the 300 hour experiment, we are going to have to work on the team report which explains our experiment, the methods, and the results. We have a notion that the Rebels team may have better results from their team experiment, so it is going to be important that we explain in detail how our experiment was run, the amount of data we used, the methods we used to run the experiment, and how that methods of running the experiment produced our results.
5/9: Touch up the final report of our experiment to make it as good as possible. Although we do want to win the competition with the other team, we still want to make sure that we present a good report to the panel of teachers who will be looking at the report. We also want to convey that we are knowledgeable with our methods of running our experiment and the results we received.
5/4: I do not really have any major concerns right now. A small concern I do have though would be that we wanted to run a baseline experiment on Idefix, to compare to the baseline experiment on Caesar. However, team rebels is still using that machine to run a 300 hour experiment and we may not have enough time by next week to get to run our 5hr experiment on Idefix.
5/6: Still no major concerns right now. It would have been nice to run another successful experiment before our final class. I would say the only concern now is if we would have time to run an experiment on the same amount of data or if we would have to run an experiment on a smaller amount of data. We were trying to stick with larger amounts of data to get a more accurate result from our experiment so my only concern would be running out of time now that the semester is almost over.
5/8: Still no major concerns at this time. The week is almost over and we were able to finish up documentation on GCC and G++, run another experiment on Idefix to compare it to results on Caesar, as well as run an experiment for the Empire group on a large amount of data and actually get back results that were a bit better than last years team. The only concern I have is how our results compared to the Rebel team and how well our method of performing an experiment compared to the Rebels method.
5/9: No concerns today. Tomorrow is the last day of class and our group has managed to do everything that we set out to do for this week. Also our team was able to run a successful experiment and we were able to write up our methods and results for our team's final report. The only concerns I have today are how the panel of judges who are going to be reading our report will like the report and how they will judge us on our experiment and results.