Scheduled Commit
Browse files
data/retrieval_individual-f36c9f97-e795-4522-8eb7-7dee254b42c7.jsonl
CHANGED
@@ -397,3 +397,6 @@
|
|
397 |
{"tstamp": 1723725599.3534, "task_type": "retrieval", "type": "chat", "model": "intfloat/multilingual-e5-large-instruct", "gen_params": {}, "start": 1723725597.1153, "finish": 1723725599.3534, "ip": "", "conv_id": "d67244c71dff47dd8578f0ae183983d6", "model_name": "intfloat/multilingual-e5-large-instruct", "prompt": "I am looking for a paper to help me deal with data constraints as I scale up my language model.", "output": [["I am looking for a paper to help me deal with data constraints as I scale up my language model.", "Title: Scaling Data-Constrained Language Models\n\nAbstract: The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations."]], "corpus": "arxiv"}
|
398 |
{"tstamp": 1723732190.2769, "task_type": "retrieval", "type": "chat", "model": "GritLM/GritLM-7B", "gen_params": {}, "start": 1723732188.0093, "finish": 1723732190.2769, "ip": "", "conv_id": "6467739d19d74bd18b68bb92509e39f4", "model_name": "GritLM/GritLM-7B", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Employee engagement\n\nEmployee engagement is a fundamental concept in the effort to understand and describe, both qualitatively and quantitatively, the nature of the relationship between an organization and its employees. An \"engaged employee\" is defined as one who is fully absorbed by and enthusiastic about their work and so takes positive action to further the organization's reputation and interests. An engaged employee has a positive attitude towards the organization and its values. In contrast, a disengaged employee may range from someone doing the bare minimum at work (aka 'coasting'), up to an employee who is actively damaging the company's work output and reputation.\nAn organization with \"high\" employee engagement might therefore be expected to outperform those with \"low\" employee engagement.\nEmployee engagement first appeared as a concept in management theory in the 1990s,\nbecoming widespread in management practice in the 2000s, but it remains contested. Despite academic critiques, employee engagement practices are well established in the management of human resources and of internal communications.\nEmployee engagement today has become synonymous with terms like 'employee experience' and 'employee satisfaction', although satisfaction is a different concept. Whereas engagement refers to work motivation, satisfaction is an employee's attitude about the job--whether they like it or not. The relevance is much more due to the vast majority of new generation professionals in the workforce who have a higher propensity to be 'distracted' and 'disengaged' at work. A recent survey by StaffConnect suggests that an overwhelming number of enterprise organizations today (74.24%) were planning to improve employee experience in 2018."]], "corpus": "wikipedia"}
|
399 |
{"tstamp": 1723732190.2769, "task_type": "retrieval", "type": "chat", "model": "BM25", "gen_params": {}, "start": 1723732188.0093, "finish": 1723732190.2769, "ip": "", "conv_id": "93a566743f014d7fabae868780558a84", "model_name": "BM25", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Diffusion of responsibility\n\nMoral disengagement Diffusion of responsibility can negatively affect personal morals. With diffusion of responsibility, it has been found that people feel less accountable for their work. This lack of accountability can be due to the fact that labour is divided amongst members in a group and so no one member feels an overwhelming amount of responsibility for their organisation or their overall project. It has been found that many members get narrowed into focusing on their individual work, that they eventually forget about any moral aspects. Purely focusing on the functional aspects of their jobs is a result of division of labour, which is a mechanism for diffusion of responsibility. This can be highly concerning for organisations since division of labour is a common practice amongst many.\nMoral disengagement is likely to be particularly important in organisations because bureaucratic structures and the division of labour seem to lend themselves to moral disengagement mechanisms such as the diffusion and displacement of responsibility. Euphemistic labelling is also common in organisations, such as when managers refer to layoffs as “rightsizing.” Also, with victims out of sight, globalization makes it easier to ignore or distort the harmful consequences of business actions. Thus, moral disengagement seems highly relevant to understanding unethical behaviour in 21st century organizations."]], "corpus": "wikipedia"}
|
|
|
|
|
|
|
|
397 |
{"tstamp": 1723725599.3534, "task_type": "retrieval", "type": "chat", "model": "intfloat/multilingual-e5-large-instruct", "gen_params": {}, "start": 1723725597.1153, "finish": 1723725599.3534, "ip": "", "conv_id": "d67244c71dff47dd8578f0ae183983d6", "model_name": "intfloat/multilingual-e5-large-instruct", "prompt": "I am looking for a paper to help me deal with data constraints as I scale up my language model.", "output": [["I am looking for a paper to help me deal with data constraints as I scale up my language model.", "Title: Scaling Data-Constrained Language Models\n\nAbstract: The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations."]], "corpus": "arxiv"}
|
398 |
{"tstamp": 1723732190.2769, "task_type": "retrieval", "type": "chat", "model": "GritLM/GritLM-7B", "gen_params": {}, "start": 1723732188.0093, "finish": 1723732190.2769, "ip": "", "conv_id": "6467739d19d74bd18b68bb92509e39f4", "model_name": "GritLM/GritLM-7B", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Employee engagement\n\nEmployee engagement is a fundamental concept in the effort to understand and describe, both qualitatively and quantitatively, the nature of the relationship between an organization and its employees. An \"engaged employee\" is defined as one who is fully absorbed by and enthusiastic about their work and so takes positive action to further the organization's reputation and interests. An engaged employee has a positive attitude towards the organization and its values. In contrast, a disengaged employee may range from someone doing the bare minimum at work (aka 'coasting'), up to an employee who is actively damaging the company's work output and reputation.\nAn organization with \"high\" employee engagement might therefore be expected to outperform those with \"low\" employee engagement.\nEmployee engagement first appeared as a concept in management theory in the 1990s,\nbecoming widespread in management practice in the 2000s, but it remains contested. Despite academic critiques, employee engagement practices are well established in the management of human resources and of internal communications.\nEmployee engagement today has become synonymous with terms like 'employee experience' and 'employee satisfaction', although satisfaction is a different concept. Whereas engagement refers to work motivation, satisfaction is an employee's attitude about the job--whether they like it or not. The relevance is much more due to the vast majority of new generation professionals in the workforce who have a higher propensity to be 'distracted' and 'disengaged' at work. A recent survey by StaffConnect suggests that an overwhelming number of enterprise organizations today (74.24%) were planning to improve employee experience in 2018."]], "corpus": "wikipedia"}
|
399 |
{"tstamp": 1723732190.2769, "task_type": "retrieval", "type": "chat", "model": "BM25", "gen_params": {}, "start": 1723732188.0093, "finish": 1723732190.2769, "ip": "", "conv_id": "93a566743f014d7fabae868780558a84", "model_name": "BM25", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Diffusion of responsibility\n\nMoral disengagement Diffusion of responsibility can negatively affect personal morals. With diffusion of responsibility, it has been found that people feel less accountable for their work. This lack of accountability can be due to the fact that labour is divided amongst members in a group and so no one member feels an overwhelming amount of responsibility for their organisation or their overall project. It has been found that many members get narrowed into focusing on their individual work, that they eventually forget about any moral aspects. Purely focusing on the functional aspects of their jobs is a result of division of labour, which is a mechanism for diffusion of responsibility. This can be highly concerning for organisations since division of labour is a common practice amongst many.\nMoral disengagement is likely to be particularly important in organisations because bureaucratic structures and the division of labour seem to lend themselves to moral disengagement mechanisms such as the diffusion and displacement of responsibility. Euphemistic labelling is also common in organisations, such as when managers refer to layoffs as “rightsizing.” Also, with victims out of sight, globalization makes it easier to ignore or distort the harmful consequences of business actions. Thus, moral disengagement seems highly relevant to understanding unethical behaviour in 21st century organizations."]], "corpus": "wikipedia"}
|
400 |
+
{"tstamp": 1723732442.2367, "task_type": "retrieval", "type": "chat", "model": "sentence-transformers/all-MiniLM-L6-v2", "gen_params": {}, "start": 1723732440.0405, "finish": 1723732442.2367, "ip": "", "conv_id": "827efe3a32744018b75b8ab8f381746b", "model_name": "sentence-transformers/all-MiniLM-L6-v2", "prompt": "what is the physical significance of dielectric constant?", "output": [["what is the physical significance of dielectric constant?", "Stackoverflow Stackexchange\n\nQ: JUnit test with dynamic number of tests In our project I have several JUnit tests that e.g. take every file from a directory and run a test on it. If I implement a testEveryFileInDirectory method in the TestCase this shows up as only one test that may fail or succeed. But I am interested in the results on each individual file. How can I write a TestCase / TestSuite such that each file shows up as a separate test e.g. in the graphical TestRunner of Eclipse? (Coding an explicit test method for each file is not an option.)\nCompare also the question ParameterizedTest with a name in Eclipse Testrunner.\n\nA: Should be possible in JUnit 3 by inheriting from TestSuite and overriding the tests() method to list the files and for each return an instance of a subclass of TestCase that takes the filename as constructor parameter and has a test method that tests the file given in the constructor.\nIn JUnit 4 it might be even easier.\n"]], "corpus": "stackexchange"}
|
401 |
+
{"tstamp": 1723732442.2367, "task_type": "retrieval", "type": "chat", "model": "BAAI/bge-large-en-v1.5", "gen_params": {}, "start": 1723732440.0405, "finish": 1723732442.2367, "ip": "", "conv_id": "753366b094894aa09f8e1861b9b6600f", "model_name": "BAAI/bge-large-en-v1.5", "prompt": "what is the physical significance of dielectric constant?", "output": [["what is the physical significance of dielectric constant?", "Physics Stackexchange\n\nQ: What is the dielectric constant of a pure conductor? Dielectric constant is the ratio of permittivity of a medium to the permittivity of free space. How to find dielectric constant of a conductor?\n\nA: Inside a metal, there is no formation of dipoles, hence there is no polarization as such. We have free electrons in metals not bound like that of a dielectric. Hence we can argue its electric susceptibility $\\chi$ = 0. We know $\\epsilon_r = 1 + \\chi$, so it can be said that its relative permittivity($\\epsilon_r$) is 1, considering electrostatics problems. For time varying fields, i.e. electrodynamics, we define complex permittivity as $\\hat{\\epsilon}=\\epsilon \\times (1+\\sigma/i\\omega\\epsilon)$, where for metals we can have the imaginary part $\\sigma/\\omega\\epsilon >> 1$. Thus for metals $\\hat{\\epsilon}=i\\sigma/\\omega)$, which is a large imaginary value considering high conductivity of metals.\nThough, not a source for this answer,the basic idea was gained by observing the value of $\\epsilon = \\epsilon_0$ being used in Introduction to Electrodynamics by David Griffiths, in problems (see Chapter 9, Problem 20,bit (b)).\n"]], "corpus": "stackexchange"}
|
402 |
+
{"tstamp": 1723732568.9323, "task_type": "retrieval", "type": "chat", "model": "BAAI/bge-large-en-v1.5", "gen_params": {}, "start": 1723732566.7731, "finish": 1723732568.9323, "ip": "", "conv_id": "e3bf323e570345359b32414cb386a565", "model_name": "BAAI/bge-large-en-v1.5", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Employee engagement\n\nEmployee engagement is a fundamental concept in the effort to understand and describe, both qualitatively and quantitatively, the nature of the relationship between an organization and its employees. An \"engaged employee\" is defined as one who is fully absorbed by and enthusiastic about their work and so takes positive action to further the organization's reputation and interests. An engaged employee has a positive attitude towards the organization and its values. In contrast, a disengaged employee may range from someone doing the bare minimum at work (aka 'coasting'), up to an employee who is actively damaging the company's work output and reputation.\nAn organization with \"high\" employee engagement might therefore be expected to outperform those with \"low\" employee engagement.\nEmployee engagement first appeared as a concept in management theory in the 1990s,\nbecoming widespread in management practice in the 2000s, but it remains contested. Despite academic critiques, employee engagement practices are well established in the management of human resources and of internal communications.\nEmployee engagement today has become synonymous with terms like 'employee experience' and 'employee satisfaction', although satisfaction is a different concept. Whereas engagement refers to work motivation, satisfaction is an employee's attitude about the job--whether they like it or not. The relevance is much more due to the vast majority of new generation professionals in the workforce who have a higher propensity to be 'distracted' and 'disengaged' at work. A recent survey by StaffConnect suggests that an overwhelming number of enterprise organizations today (74.24%) were planning to improve employee experience in 2018."]], "corpus": "wikipedia"}
|
data/retrieval_single_choice-f36c9f97-e795-4522-8eb7-7dee254b42c7.jsonl
CHANGED
@@ -1 +1,2 @@
|
|
1 |
{"tstamp": 1723685045.5241, "task_type": "retrieval", "type": "upvote", "models": "Alibaba-NLP/gte-Qwen2-7B-instruct", "ip": "", "conv_id": "dbe077a2e59c42d7ad8b8a6d33f0b0c0", "model_name": "Alibaba-NLP/gte-Qwen2-7B-instruct", "prompt": "解釋便利商店的職責", "output": [["解釋便利商店的職責", "Convenience store\n\nA convenience store, convenience shop, bodega, corner store or corner shop is a small retail store that stocks a range of everyday items such as tea, coffee, groceries, fruits, vegetables, snacks, confectionery, soft drinks, ice creams, tobacco products, lottery tickets, over-the-counter drugs, toiletries, newspapers and magazines.\nIn some jurisdictions, convenience stores are licensed to sell alcoholic drinks, although many jurisdictions limit such beverages to those with relatively low alcohol content, like beer and wine. The stores may also offer money order and wire transfer services, along with the use of a fax machine or photocopier for a small per-copy cost. Some also sell tickets or recharge smart cards, e.g. OPUS cards in Montreal, Canada or include a small deli. They differ from general stores and village shops in that they are not in a rural location and are used as a convenient (hence their common name) supplement to larger stores.\nA convenience store may be part of a gas/petrol station, so customers can purchase goods while refuelling their vehicle. It may be located alongside a busy road, in an urban area, near a railway or railroad station or other transport hub. In some countries, convenience stores have long shopping hours and some remain open 24 hours."]], "corpus": "wikipedia"}
|
|
|
|
1 |
{"tstamp": 1723685045.5241, "task_type": "retrieval", "type": "upvote", "models": "Alibaba-NLP/gte-Qwen2-7B-instruct", "ip": "", "conv_id": "dbe077a2e59c42d7ad8b8a6d33f0b0c0", "model_name": "Alibaba-NLP/gte-Qwen2-7B-instruct", "prompt": "解釋便利商店的職責", "output": [["解釋便利商店的職責", "Convenience store\n\nA convenience store, convenience shop, bodega, corner store or corner shop is a small retail store that stocks a range of everyday items such as tea, coffee, groceries, fruits, vegetables, snacks, confectionery, soft drinks, ice creams, tobacco products, lottery tickets, over-the-counter drugs, toiletries, newspapers and magazines.\nIn some jurisdictions, convenience stores are licensed to sell alcoholic drinks, although many jurisdictions limit such beverages to those with relatively low alcohol content, like beer and wine. The stores may also offer money order and wire transfer services, along with the use of a fax machine or photocopier for a small per-copy cost. Some also sell tickets or recharge smart cards, e.g. OPUS cards in Montreal, Canada or include a small deli. They differ from general stores and village shops in that they are not in a rural location and are used as a convenient (hence their common name) supplement to larger stores.\nA convenience store may be part of a gas/petrol station, so customers can purchase goods while refuelling their vehicle. It may be located alongside a busy road, in an urban area, near a railway or railroad station or other transport hub. In some countries, convenience stores have long shopping hours and some remain open 24 hours."]], "corpus": "wikipedia"}
|
2 |
+
{"tstamp": 1723732580.4001, "task_type": "retrieval", "type": "flag", "models": "BAAI/bge-large-en-v1.5", "ip": "", "conv_id": "e3bf323e570345359b32414cb386a565", "model_name": "BAAI/bge-large-en-v1.5", "prompt": "Employee disengagement in an organisation", "output": [["Employee disengagement in an organisation", "Employee engagement\n\nEmployee engagement is a fundamental concept in the effort to understand and describe, both qualitatively and quantitatively, the nature of the relationship between an organization and its employees. An \"engaged employee\" is defined as one who is fully absorbed by and enthusiastic about their work and so takes positive action to further the organization's reputation and interests. An engaged employee has a positive attitude towards the organization and its values. In contrast, a disengaged employee may range from someone doing the bare minimum at work (aka 'coasting'), up to an employee who is actively damaging the company's work output and reputation.\nAn organization with \"high\" employee engagement might therefore be expected to outperform those with \"low\" employee engagement.\nEmployee engagement first appeared as a concept in management theory in the 1990s,\nbecoming widespread in management practice in the 2000s, but it remains contested. Despite academic critiques, employee engagement practices are well established in the management of human resources and of internal communications.\nEmployee engagement today has become synonymous with terms like 'employee experience' and 'employee satisfaction', although satisfaction is a different concept. Whereas engagement refers to work motivation, satisfaction is an employee's attitude about the job--whether they like it or not. The relevance is much more due to the vast majority of new generation professionals in the workforce who have a higher propensity to be 'distracted' and 'disengaged' at work. A recent survey by StaffConnect suggests that an overwhelming number of enterprise organizations today (74.24%) were planning to improve employee experience in 2018."]], "corpus": "wikipedia"}
|