init

.gitignore

@@ -0,0 +1,162 @@
+# 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/
+
+*.safetensors

README.md

@@ -1,10 +1,11 @@
 ---
-title: Rwkv Rs
+title: RWKV-rs 7b
 emoji: 🐢
 colorFrom: purple
 colorTo: green
 sdk: gradio
 sdk_version: 3.18.0
+python_version: 3.10
 app_file: app.py
 pinned: false
 license: agpl-3.0

app.py

@@ -0,0 +1,104 @@
+import os
+
+import rwkv_rs
+import numpy as np
+import huggingface_hub
+import tokenizers
+
+import gradio as gr
+
+model_path = "./rnn.safetensors"
+if not os.path.exists(model_path):
+    model_path = huggingface_hub.hf_hub_download(repo_id="mrsteyk/RWKV-LM-safetensors", filename="RWKV-4-Pile-7B-Instruct-test1-20230124.rnn.safetensors")
+assert model_path is not None
+
+model = rwkv_rs.Rwkv(model_path)
+tokenizer = tokenizers.Tokenizer.from_pretrained("EleutherAI/gpt-neox-20b")
+
+GT = [
+    gr.Button.update(visible=False),
+    gr.Button.update(visible=True),
+]
+GF = [
+    gr.Button.update(visible=True),
+    gr.Button.update(visible=False),
+]
+
+def complete_fn(inpt, max_tokens, min_tokens, alpha_f, alpha_p):
+    try:
+        state = rwkv_rs.State(model)
+        text = inpt
+        counts = [0]*tokenizer.get_vocab_size()
+        tokens = tokenizer.encode(inpt).ids
+        # yield ("Preproc...", gr.Text.update(visible=False))
+        # logits = model.forward(tokens, state)
+        for i in range(len(tokens) - 1):
+            model.forward_token_preproc(tokens[i], state)
+            yield (tokenizer.decode(tokens[:i + 1]), None)
+        logits = model.forward_token(tokens, state)
+        yield (text, None)
+        max_tokens = int(max_tokens)
+        for i in range(max_tokens):
+            if i < min_tokens:
+                logits[0] = -100
+            for i in range(len(counts)):
+                logits[i] -= (counts[i]* alpha_f) + (float(counts[i] > 0) * alpha_p)
+            token = np.argmax(logits)
+            counts[token] += 1
+            if token == 0:
+                break
+            if i == max_tokens - 1:
+                break
+            tokens += [token]
+            text = tokenizer.decode(tokens)
+            yield (text, None)
+            logits = model.forward_token(token, state)
+        yield (text, None)
+    except Exception as e:
+        print(e)
+        yield ("Error...", gr.Text.update(value=str(e), visible=True))
+    # finally:
+    #     return (None, None)
+
+def generator_wrap(l, fn):
+    def wrap(*args):
+        last_i = list([None] * l)
+        try:
+            for i in fn(*args):
+                last_i = list(i)
+                yield last_i + GT
+        finally:
+            yield last_i + GF
+    return wrap
+
+
+with gr.Blocks() as app:
+    gr.Markdown(f"Running on `{model_path}`")
+    error_box = gr.Text(label="Error", visible=False)
+
+    with gr.Tab("Complete"):
+        with gr.Row():
+            inpt = gr.TextArea(label="Input")
+            out = gr.TextArea(label="Output")
+        complete = gr.Button("Complete", variant="primary")
+        c_stop = gr.Button("Stop", variant="stop", visible=False)
+    with gr.Tab("Insert (WIP)"):
+        gr.Markdown("WIP, use `<|INSERT|>` to indicate a place to replace")
+        with gr.Row():
+            inpt_i = gr.TextArea(label="Input")
+            out_i = gr.TextArea(label="Output")
+        insert = gr.Button("Insert")
+
+    with gr.Column():
+        max_tokens = gr.Slider(label="Max Tokens", minimum=1, maximum=4096, step=1, value=767)
+        min_tokens = gr.Slider(label="Min Tokens", minimum=0, maximum=4096, step=1)
+        alpha_f = gr.Slider(label="Alpha Frequency", minimum=0, maximum=100, step=0.01)
+        alpha_p = gr.Slider(label="Alpha Presence", minimum=0, maximum=100, step=0.01)
+
+    G = [complete, c_stop]
+
+    c = complete.click(generator_wrap(2, complete_fn), [inpt, max_tokens, min_tokens, alpha_f, alpha_p], [out, error_box] + G)
+    c_stop.click(lambda: (complete.update(visible=True), c_stop.update(visible=False)), inputs=None, outputs=[complete, c_stop], cancels=[c], queue=False)
+
+app.queue()
+app.launch()

requirements.txt

@@ -0,0 +1,5 @@
+gradio==3.18.0
+huggingface_hub
+tokenizers
+numpy
+rwkv_rs>=0.2.0, <0.3.0