feat: Model links updated for first public release

README.md

@@ -1,16 +1,11 @@
 # LibertAI Agents
 
-## Supported models
+LibertAI Agents are highly customizable assistants that operates in a decentralized and confidential environment.\
+They can be given tools to perform tasks (search something on the web, make a transaction on-chain...), making them
+powerful companions.
 
-We support multiple open-source models that have agentic capabilities.
+> 💡 Agents are currently in private beta, contact the team [on Telegram](https://t.me/libertai) if you
+want to try it out!\
+We are planning some improvements and will listen to beta tester's feedbacks before releasing a stable version.
 
-- [Hermes 2 Pro - Llama 3 8B](https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B)
-- ⏳ [Hermes 3 - Llama-3.1 8B](https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B)
-- ⏳ [Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407)
-
-## Using a gated model
-
-Some models, like [Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) are gated (
-generally to require you to accept some usage conditions).\
-To use those models, you need to create an [access token](https://huggingface.co/settings/tokens) from your Hugging Face
-account and give it to the `get_model` function.
+This repository contains the source code of the actual [Python framework](https://pypi.org/project/libertai-agents) that you can use to create your agents, along with the backend used internally to manage their deployment on [Aleph.im](https://aleph.im)'s decentralized and confidential VMs.

backend/.env.example

@@ -1,10 +1,10 @@
 # Use the testnet in development
 # ALEPH_API_URL=https://api.twentysix.testnet.network
 
-# Sender of the messages on Aleph
-ALEPH_SENDER=0x00
 # Owner of the messages on Aleph (has given permissions to the sender)
 ALEPH_OWNER=0x00
+# Sender of the messages on Aleph
+ALEPH_SENDER=0x00
 # Private key of the address to send messages with
 ALEPH_SENDER_SK=
 # Public key of the address to send messages with

libertai_agents/.gitignore

@@ -1,2 +1,5 @@
 /venv
-/dist
+/dist
+
+# Development test file
+/dev.py

libertai_agents/README.md

@@ -1,16 +1,7 @@
 # LibertAI Agents
 
-## Supported models
+Framework to create and deploy decentralized agents 🚀
 
-We support multiple open-source models that have agentic capabilities.
+> ⚠ This framework is in beta and might undergo some breaking changes before the stable release.
 
-- [Hermes 2 Pro - Llama 3 8B](https://huggingface.co/NousResearch/Hermes-2-Pro-Llama-3-8B)
-- ⏳ [Hermes 3 - Llama-3.1 8B](https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B)
-- ⏳ [Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407)
-
-## Using a gated model
-
-Some models, like [Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407) are gated (
-generally to require you to accept some usage conditions).\
-To use those models, you need to create an [access token](https://huggingface.co/settings/tokens) from your Hugging Face
-account and give it to the `get_model` function.
+Take a look at [our documentation](https://docs.libertai.io/agents) to get started!

libertai_agents/libertai_agents/models/base.py

@@ -8,7 +8,6 @@
 logging.getLogger("transformers").disabled = True
 
 ModelId = Literal[
-    "NousResearch/Hermes-2-Pro-Llama-3-8B",
     "NousResearch/Hermes-3-Llama-3.1-8B",
     "mistralai/Mistral-Nemo-Instruct-2407"
 ]

libertai_agents/libertai_agents/models/models.py

@@ -16,17 +16,12 @@ class ModelConfiguration(BaseModel):
 
 MODEL_IDS: list[ModelId] = list(typing.get_args(ModelId))
 
-# TODO: update URLs with prod, and check context size (if we deploy it with a lower one)
 MODELS_CONFIG: dict[ModelId, ModelConfiguration] = {
-    "NousResearch/Hermes-2-Pro-Llama-3-8B": ModelConfiguration(
-        vm_url="https://curated.aleph.cloud/vm/84df52ac4466d121ef3bb409bb14f315de7be4ce600e8948d71df6485aa5bcc3/completion",
-        context_length=8192,
-        constructor=HermesModel),
-    "NousResearch/Hermes-3-Llama-3.1-8B": ModelConfiguration(vm_url="http://localhost:8080/completion",
-                                                             context_length=131_072,
+    "NousResearch/Hermes-3-Llama-3.1-8B": ModelConfiguration(vm_url="https://curated.aleph.cloud/vm/84df52ac4466d121ef3bb409bb14f315de7be4ce600e8948d71df6485aa5bcc3/completion",
+                                                             context_length=4096,
                                                              constructor=HermesModel),
-    "mistralai/Mistral-Nemo-Instruct-2407": ModelConfiguration(vm_url="http://localhost:8080/completion",
-                                                               context_length=131_072,
+    "mistralai/Mistral-Nemo-Instruct-2407": ModelConfiguration(vm_url="https://curated.aleph.cloud/vm/2c4ad0bf343fb12924936cbc801732d95ce90f84cd895aa8bee82c0a062815c2/completion",
+                                                               context_length=4096,
                                                                constructor=MistralModel)
 }
 

libertai_agents/pyproject.toml

@@ -1,12 +1,12 @@
 [tool.poetry]
 name = "libertai-agents"
-version = "0.0.18"
-description = ""
+version = "0.0.1"
+description = "Framework to create and deploy decentralized agents"
 authors = ["LibertAI.io team <[email protected]>"]
 readme = "README.md"
 homepage = "https://libertai.io"
 repository = "https://github.com/LibertAI/libertai-agents"
-documentation = "https://docs.libertai.io"
+documentation = "https://docs.libertai.io/agents"
 classifiers = [
     "Operating System :: OS Independent",
     "Intended Audience :: Developers",