Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Add skill for creating tables in Asciidoc format (instructlab#734)
This is a re-submission of instructlab#187. If your PR is related to a contribution to the taxonomy, please, fill out the following questionnaire. If not, replace this whole text and the following questionnaire with whatever information is applicable to your PR. **Describe the contribution to the taxonomy** <!-- A concise description of what the contribution brings, replace "..." in the bullet list --> - Currently the merlinite model generates a markdown table when asked to generate an asciidoc table. This skill attempts to teach it how to generate Asciidoc tables. - I'm not sure about the path I've put the qna.yaml in here, please let me know if it should be somewhere else. I'll give several examples below of various questions asked before and after, and will also include an image showing the before and after rendered in an HTML page that was rendered from trying to interpret the responses as Asciidoc. **Try it out yourself!** I've put the trained model from `lab train` onto Google Drive, so that you can download it and ask your own questions, and not just rely on the examples that I've provided below. You can find it here (note that it's 13.5GB): https://drive.google.com/file/d/12JJhsZPWQxXZb-WZZUof_Cit2t09Y84n/view?usp=sharing **Input given at the prompt** <!-- What you entered, replace "..." --> _See examples below the "Contribution checklist" section. I've put them at the end because some of the Markdown features I've used are a bit odd, and can cause effects to sections following them._ **Response from the original model** <!-- What you received from the original model in response to your input, replace "..." --> _See examples below the "Contribution checklist" section. I've put them at the end because some of the Markdown features I've used are a bit odd, and can cause effects to sections following them._ **Response from the fine-tuned model** <!-- Generate a synthetic dataset based on your newly added seed data; train the model with the synthetic data and now re-test the model's response with the same prompt. Replace "..." with what you receive with the finetuned model. --> _See examples below the "Contribution checklist" section. I've put them at the end because some of the Markdown features I've used are a bit odd, and can cause effects to sections following them._ **Contribution checklist** <!-- Insert an x between the empty brackets: [ ] >> [x] --> - [x] The contribution was tested with `ilab generate` - [x] No errors or warnings were produced by `ilab generate` - [x] All [commits are signed off](https://github.com/instructlab/taxonomy/blob/main/CONTRIBUTING.md#legal) (DCO) - [x] The `qna.yaml` file contains at least 5 `seed_examples` - [x] The `qna.yaml` file was [linted](https://yamllint.com) and [prettified](https://onlineyamltools.com/prettify-yaml) ([yaml-validator](https://jsonformatter.org/yaml-validator) can do both) - [x] An `attribution.txt` file in the same folder as the `qna.yaml` file. ---- ### Example 1 > Create an AsciiDoc table showing the IMDB top 5 movies, including the year they were released, and their IMDB rating <details> <summary>Response that was received</summary> <pre><code> .. csv-table:: Top 5 IMDb Movies of All Time :header: Title, Year, IMDb Rating "The Godfather", 1972, 9.2 "The Shawshank Redemption", 1994, 9.3 "The Godfather: Part II", 1974, 9.0 "Pulp Fiction", 1994, 8.9 "The Dark Knight", 2008, 9.0 </code></pre> </details> <details> <summary>Response that is now received instead</summary> <pre><code> Here's the AsciiDoc code for the table: .IMDB Top 5 Movies [%header] |=== | Movie | Year | Rating | The Godfather | 1972 | 9.2 | The Shawshank Redemption | 1994 | 9.3 | The Dark Knight | 2008 | 9.0 | Pulp Fiction | 1994 | 8.9 | Forrest Gump | 1994 | 8.8 |=== The rendered HTML output will look like this: IMDB Top 5 Movies ----------------- | Movie | Year | Rating | The Godfather | 1972 | 9.2 | The Shawshank Redemption | 1994 | 9.3 | The Dark Knight | 2008 | 9.0 | Pulp Fiction | 1994 | 8.9 | Forrest Gump | 1994 | 8.8 </code></pre> </details> <details> <summary>Example screenshot of before and after as rendered HTML from an Asciidoc processor on the responses</summary> <img src="https://github.com/instruct-lab/taxonomy/assets/442386/77d35479-3b82-41a0-9e0c-c7418f127bb1"> </details> ### Example 2 > Create an Asciidoc-formatted table with 4 rows and two columns, where the first column is a random letter, and the second column is a random number between 1 and 4. <details> <summary>Response that was received</summary> <pre><code> Certainly! I can help you create a randomized AsciiDoc table with the specified structure. However, please note that I am an AI language model and do not have the ability to generate truly random values. I will provide you with a set of random values that you can use for the table. Here's an example of how the table might look: ``` = Table: Random Letters and Numbers | Random Letter | Random Number | | --- | --- | | A | 1 | | B | 2 | | C | 3 | | D | 4 | ``` Feel free to replace the provided values with your own random letters and numbers. If you need help generating random values, you can use an online random number generator or a programming language to create the values for the table. </code></pre> </details> <details> <summary>Response that is now received instead</summary> <pre><code> Here's the AsciiDoc code for the table: .Table Example [%header] |=== | Column 1 | Column 2 | A | 1 | B | 2 | C | 3 | D | 4 |=== The rendered HTML output will look like this: Table Example ------------- | Column 1 | Column 2 | A | 1 | B | 2 | C | 3 | D | 4 </code></pre> </details> <details> <summary>Example screenshot of before and after as rendered HTML from an Asciidoc processor on the responses</summary> <img src="https://github.com/instruct-lab/taxonomy/assets/442386/f4352bb1-fa78-4169-9756-2ccff0d8db01"> </details> ### Example 3 > Convert the following Markdown table to AsciiDoc: > > | First name | Last name | Age | Score | > | ---------- | --------- | --- | ----- | > | Alice | O'Wonderland | 20 | 53 | > | Bob | Tables | 10 | null | > | Carol | Basking | 60 | 25 | > | Dave | Letterbox | 55 | 40 | <details> <summary>Response that was received</summary> <pre><code> Here's the AsciiDoc version of the table: .. csv-table:: Table: Users' Information :header-rows: 1 "First name", "Last name", "Age", "Score" "Alice", "O'Wonderland", "20", "53" "Bob", "Tables", "10", "null" "Carol", "Basking", "60", "25" "Dave", "Letterbox", "55", "40" The table is created using the ``csv-table`` directive, which is suitable for simple tables like this one. </code></pre> </details> <details> <summary>Response that is now received instead</summary> <pre><code> Here is the equivalent AsciiDoc equivalent: .Table: Students' Information [%header] |=== | First Name | Last Name | Age | Score | ----------- | ------------ | --- | ----- | Alice | O'Wonderland | 20 | 53 | Bob | Tables | 10 | null | Carol | Basking | 60 | 25 | Dave | Letterbox | 55 | 40 |=== Note that the header row is formatted in brackets, and the rows that follow it use the "|" delimiter to indicate the column contents. The table is surrounded by "|===" lines to define the table boundaries. </code></pre> </details> <details> <summary>Example screenshot of before and after as rendered HTML from an Asciidoc processor on the responses</summary> <img src="https://github.com/instruct-lab/taxonomy/assets/442386/28d56989-76a7-42f6-9946-72b9e96fbdf1"> </details> ### Example 4 > Given the following csv file contents, how could it be represented as an asciidoc table? 1,2,3,4 5,6,7,8 9,10,11,12 <details> <summary>Response that was received</summary> <pre><code> To represent the given CSV file contents as an ASCIIDOC table, you can use the `table()` directive along with the `row()` macro. Here's a sample ASCIIDOC code that creates a table from the given CSV data: =table ---- :name: table-csv :source: csv-file.csv :header: 1,2,3,4 =table ---- :: :source: table-csv ::row() 1,2,3,4 ::row() 5,6,7,8 ::row() 9,10,11,12 In this example, the `table()` directive is used to define a table with a given width for each column. The `:source` attribute is used to specify the name of the CSV file that contains the data for the table. The `:header` attribute is used to specify the header row of the CSV file. The `row()` macro is then used to add rows to the table. This will result in the following table: | 1 | 2 | 3 | 4 | | - | - | - | - | | 5 | 6 | 7 | 8 | | 9 | 10 | 11 | 12 | </code></pre> </details> <details> <summary>Response that is now received instead</summary> <pre><code> Here's the asciidoc code for the table: .Table: CSV Data [%header] |=== | Column 1 | Column 2 | Column 3 | Column 4 |----------|----------|----------|---------- | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |=== The asciidoc table is now represented using the .Table: CSV Data header, and the rows and columns are defined using the | row/column label | data coordinates format. The data coordinates indicate the position of the data in the table. </code></pre> </details> <details> <summary>Example screenshot of before and after as rendered HTML from an Asciidoc processor on the responses</summary> <img src="https://github.com/instruct-lab/taxonomy/assets/442386/b127929f-5705-4049-84d9-ee9ff7e01bf4"> </details> ### Example 5 > What is the syntax for creating an Asciidoc table? <details> <summary>Response that was received</summary> <pre><code> In Asciidoc, you can create a table using the `table` directive, followed by rows and columns defined using pipelines (`|`) to separate cells. Here's a simple example of an Ascidoc table: ``` =table | row1col1, row1col2, row1col3, row1col4, row1col5 | row2col1, row2col2, row2col3, row2col4, row2col5 | ...] ``` This creates a table with five headers and multiple rows. Each row contains five cells separated by pipelines. If you want to create a simple two-column table, you can use this syntax: ``` =table | row1col1, row1col2 | row2col1, row2col2 | ...] ``` In this example, `header1` and `header2` are the headers of the two columns, and `row1col1`, `row1col2`, `row2col1`, and `row2col2` are the cells of the two rows. You can replace these placeholders with your desired content. Keep in mind that Asciidoc tables support various formatting options, such as aligning cells, merging cells, and using different table styles. For more information on these features, refer to the official Asciidoctor documentation or use a search engine to find additional resources on Asciidoctor tables. </code></pre> </details> <details> <summary>Response that is now received instead</summary> <pre><code> Here's the syntax for creating an Asciidoc table: .. Table: Example Table - Asciidoc Syntax :name: ascii-table-syntax :widths: 20, 80 |=== | Column1 | Column2 | |---------|---------| | Value1 | Content | | Value2 | Of the | | | second | | | column | |---------|---------| |=== The table has the following structure: - The first line starts with the "|===" delimiter, followed by the table name, widths, and widths parameters (if needed). - Each subsequent line starts with the "|" delimiter, followed by the column labels or values. - The table ends with the "|===" delimiter. The widths parameter is used to specify the width of each column in terms of the number of characters. If not provided, the widths will be automatically calculated based on the longest label or value for each column. Here's the AsciiDoc code for the example table: .Table: AsciiDoc Table Syntax [%header] :name: ascii-table-syntax :widths: 20, 80 |=== | Column1 | Column2 | |---------|---------| | Value1 | Content | | Value2 | Of the | | | second | | | column | |---------|---------| |=== The rendered HTML output will display the table with the given structure and content. </code></pre> </details> <details> <summary>Example screenshot of before and after as rendered HTML from an Asciidoc processor on the responses</summary> <img src="https://github.com/instruct-lab/taxonomy/assets/442386/ff21bbcf-54b9-4eed-895e-714ef2eb4492"> </details> --------- Signed-off-by: Gerard Ryan <git@grdryn.xyz> Signed-off-by: BJ Hargrave <hargrave@us.ibm.com>
- Loading branch information
