init function calling

Signed-off-by: SumanthRH <sumanthrh@anyscale.com>

.gitignore

@@ -0,0 +1,166 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# jobs output
+./jobs
+
+# Vim
+*.swp

.pre-commit-config.yaml

@@ -4,6 +4,7 @@ repos:
     rev: v3.2.0
     hooks:
       - id: check-added-large-files
+        args: ['--maxkb=10000']
       - id: trailing-whitespace
         # README might be auto-generated
         exclude: templates/.+/README.md

templates/fine-tune-function-calling/fc_utils/data_format_checker.py

@@ -0,0 +1,48 @@
+"""
+Data format validation function from https://docs.endpoints.anyscale.com/endpoints/fine-tuning/dataset-prep/#context-length
+"""
+
+
+class FormattingError(Exception):
+    pass
+
+
+def convert_message_list_to_text(messages: list) -> str:
+    # special tokens are for the Llama-2 model
+    B_INST, E_INST = "[INST]", "[/INST]"
+    B_SYS, E_SYS = "<<SYS>>\n", "\n<</SYS>>\n\n"
+    text = ""
+
+    if messages[0]["role"] == "system":
+        messages = [
+            {
+                "role": messages[1]["role"],
+                "content": B_SYS
+                + messages[0]["content"]
+                + E_SYS
+                + messages[1]["content"],
+            }
+        ] + messages[2:]
+
+    assert all([msg["role"] == "user" for msg in messages[::2]]) and all(
+        [msg["role"] == "assistant" for msg in messages[1::2]]
+    ), (
+        "model only supports 'system','user' and 'assistant' roles, "
+        "starting with user and alternating (u/a/u/a/u...)"
+    )
+
+    texts = []
+    for prompt, answer in zip(messages[::2], messages[1::2]):
+        texts.append(
+            f"{B_INST} {(prompt['content']).strip()} {E_INST} {(answer['content']).strip()} "
+        )
+
+    text = "</s><s>".join(texts)
+    # add the bos and eos token at the beginning of the first turn and the end of the last turn
+    text = "<s>" + text + " </s>"
+    # During training last message should be from assistant (not from a user)
+    assert (
+        messages[-1]["role"] == "assistant"
+    ), f"Last message must be from assistant, got {messages[-1]['role']}"
+
+    return text

templates/fine-tune-function-calling/fc_utils/data_viz.py

@@ -0,0 +1,141 @@
+import datasets
+import matplotlib.pyplot as plt
+import numpy as np
+import ray.data
+
+from fc_utils.preprocessing import initial_mapper, pprint_example
+
+hf_ds = datasets.load_dataset(
+    "glaiveai/glaive-function-calling-v2", split="train"
+).shuffle()
+train_hf_ds = hf_ds.select(range(int(len(hf_ds) * 0.1)))
+test_hf_ds = hf_ds.select(range(int(len(hf_ds) * 0.1), int(len(hf_ds) * 0.11)))
+ray_ds = ray.data.from_huggingface(hf_ds)
+openai_fmt_ds = ray_ds.map(initial_mapper)
+
+pprint_example(openai_fmt_ds.take(1)[0])
+
+
+def get_counts(ds):
+    counts = {
+        "Normal-Single": [0, 0, 0],
+        "Normal-Multi": [0, 0, 0],
+        "Tool-Single": [0, 0, 0],
+        "Tool-Multi": [0, 0, 0],
+    }
+    for ex in ds.iter_rows():
+        count = len(eval(ex["tools"]) if ex["tools"] else [])
+        roles = [message["role"] for message in ex["messages"]]
+        num_user = len([role for role in roles if role == "user"])
+        is_multi = num_user > 1 if roles[-1] != "user" else num_user > 2
+        if count == 0:
+            if is_multi:
+                counts["Normal-Multi"][0] += 1
+            else:
+                counts["Normal-Single"][0] += 1
+        elif count == 1:
+            if "tool" not in roles:
+                if is_multi:
+                    counts["Normal-Multi"][1] += 1
+                else:
+                    counts["Normal-Single"][1] += 1
+            else:
+                if is_multi:
+                    counts["Tool-Multi"][1] += 1
+                else:
+                    counts["Tool-Single"][1] += 1
+        else:
+            if "tool" not in roles:
+                if is_multi:
+                    counts["Normal-Multi"][2] += 1
+                else:
+                    counts["Normal-Single"][2] += 1
+            else:
+                if is_multi:
+                    counts["Tool-Multi"][2] += 1
+                else:
+                    counts["Tool-Single"][2] += 1
+    return counts
+
+
+counts = get_counts(openai_fmt_ds)
+
+
+# Number of bars per group
+n_groups = 3
+fig, ax = plt.subplots()
+index = np.arange(n_groups)
+bar_width = 0.2
+opacity = 0.8
+
+total_count = openai_fmt_ds.count()
+normal_single = 100 * np.array(counts["Normal-Single"]) / total_count
+normal_multi = 100 * np.array(counts["Normal-Multi"]) / total_count
+tool_single = 100 * np.array(counts["Tool-Single"]) / total_count
+tool_multi = 100 * np.array(counts["Tool-Multi"]) / total_count
+
+# code for pie chart version
+# labels = ['0 Funcs; Normal Response; Single Turn', '0 Funcs; Normal Response; Multi Turn', '1 Func; Normal Response; Single Turn', '1 Func; Normal Response; Multi Turn', '1 Func; Tool Call; Single Turn', '1 Func; Tool Call; Multi Turn', '2 Funcs; Tool Call; Single Turn', '2 Funcs; Tool Call; Multi Turn']
+# sizes = [normal_single[0], normal_multi[0], normal_single[1], normal_multi[1], tool_single[1], tool_multi[1], tool_single[2], tool_multi[2]]
+
+# fig, ax = plt.subplots()
+# ax.pie(sizes, labels=labels, autopct='%1.2f%%')
+
+rects1 = plt.bar(
+    index, normal_single, bar_width, alpha=opacity, color="b", label="Normal-Single"
+)
+rects2 = plt.bar(
+    index + bar_width,
+    normal_multi,
+    bar_width,
+    alpha=opacity,
+    color="g",
+    label="Normal-Multi",
+)
+rects3 = plt.bar(
+    index + 2 * bar_width,
+    tool_single,
+    bar_width,
+    alpha=opacity,
+    color="r",
+    label="Tool-Single",
+)
+rects4 = plt.bar(
+    index + 3 * bar_width,
+    tool_multi,
+    bar_width,
+    alpha=opacity,
+    color="y",
+    label="Tool-Multi",
+)
+
+
+# Adding percentages on top of the bars
+def add_labels(rects):
+    for rect in rects:
+        height = rect.get_height()
+        ax.annotate(
+            f"{height:.1f}%",
+            xy=(rect.get_x() + rect.get_width() / 2, height),
+            xytext=(0, 3),
+            textcoords="offset points",
+            ha="center",
+            va="bottom",
+            fontsize=10,
+        )
+
+
+for rect in [rects1, rects2, rects3, rects4]:
+    add_labels(rect)
+
+# add_labels([rects1, rects2, rects3, rects4])
+
+
+plt.xlabel("Number of Tools")
+plt.ylabel("Percentage")
+plt.title("Distribution of Responses and Turns with Number of Tools")
+plt.xticks(index + bar_width * 1.5, [0, 1, 2])
+plt.legend()
+
+plt.tight_layout()
+plt.show()