Instructions to use SiddardhaShayini/TAEL_Agent with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use SiddardhaShayini/TAEL_Agent with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf SiddardhaShayini/TAEL_Agent:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf SiddardhaShayini/TAEL_Agent:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf SiddardhaShayini/TAEL_Agent:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf SiddardhaShayini/TAEL_Agent:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Use Docker

docker model run hf.co/SiddardhaShayini/TAEL_Agent:Q4_K_M

LM Studio
Jan

vLLM

How to use SiddardhaShayini/TAEL_Agent with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SiddardhaShayini/TAEL_Agent"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SiddardhaShayini/TAEL_Agent",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SiddardhaShayini/TAEL_Agent:Q4_K_M

Ollama
How to use SiddardhaShayini/TAEL_Agent with Ollama:
```
ollama run hf.co/SiddardhaShayini/TAEL_Agent:Q4_K_M
```
Unsloth Desktop

How to use SiddardhaShayini/TAEL_Agent with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "SiddardhaShayini/TAEL_Agent:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use SiddardhaShayini/TAEL_Agent with Docker Model Runner:
```
docker model run hf.co/SiddardhaShayini/TAEL_Agent:Q4_K_M
```

Lemonade

How to use SiddardhaShayini/TAEL_Agent with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull SiddardhaShayini/TAEL_Agent:Q4_K_M

Run and chat with the model

lemonade run user.TAEL_Agent-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use SiddardhaShayini/TAEL_Agent with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default SiddardhaShayini/TAEL_Agent:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use SiddardhaShayini/TAEL_Agent with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf SiddardhaShayini/TAEL_Agent:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "SiddardhaShayini/TAEL_Agent:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

TAEL-Agent (`v2.0.0-rc1`) — Granite 3.1 2B Instruct GGUF (`Q4_K_M`)

GitHub Repository (Agent Runtime, Tool System, Memory & VS Code Extension): https://github.com/SiddardhaShayini/TAEL_Agent
Hugging Face Model Repository: https://huggingface.co/SiddardhaShayini/TAEL_Agent
Base Model: ibm-granite/granite-3.1-2b-instruct
Quantization: Q4_K_M (1,473.71 MB / 1.44 GiB on disk)
License: Apache 2.0

TAEL-Agent (Task-Adaptive Efficient Language) is an autonomous software engineering and coding AI model engineered to run smoothly on 8GB RAM consumer laptops without a dedicated GPU (14–23 tokens/sec on CPU AVX2 threads).

1. Repository Files (`SiddardhaShayini/TAEL_Agent`)

File	Size	Description
`granite-3.1-2b-instruct.Q4_K_M.gguf`	`1.47 GB`	4-bit `Q4_K_M` quantized GGUF v3 model (`362` tensors, `36` KV metadata entries).
`Modelfile`	`1.0 KB`	Production Ollama `Modelfile` with ChatML template (`<
`config.json`	`850 B`	Model architecture configuration (`granite`, 32 layers, 2048 hidden size).
`README.md`	`6.5 KB`	This Hugging Face Model Card.

2. Quickstart with Ollama

# 1. Download granite-3.1-2b-instruct.Q4_K_M.gguf and Modelfile into the same folder
ollama create tael-agent -f Modelfile

# 2. Run interactive session
ollama run tael-agent

3. Quickstart with `llama-cpp-python`

from llama_cpp import Llama

llm = Llama(
    model_path="granite-3.1-2b-instruct.Q4_K_M.gguf",
    n_ctx=4096,
    n_threads=6,
    verbose=False,
)

resp = llm.create_chat_completion(
    messages=[
        {"role": "system", "content": "You are TAEL, an autonomous software engineering AI assistant."},
        {"role": "user", "content": "Write an O(n) Python function for Two Sum."}
    ],
    temperature=0.15,
    max_tokens=512,
)
print(resp["choices"][0]["message"]["content"])

4. 3-Phase Progressive Curriculum Fine-Tuning (`29,900` Samples)

Trained with Unsloth FastLanguageModel (QLoRA) ($r=16, \alpha=32$) across all projection modules (q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj) with progressively decreasing learning rates:

Phase 1 — Foundation (18,000 samples, lr = 2e-4): Programming (15k), Software Architecture (1k), Unit Testing (2k).
Phase 2 — Refinement (4,400 samples, lr = 1.5e-4): Debugging (2.4k) and Guard-Clause Refactoring (2.0k).
Phase 3 — Agentic SWE (7,500 samples, lr = 1e-4): Multi-step Planning (2k), Tool Calling (2k), Terminal (1k), Git (1k), SWE-Agent (1k), Repository Context (500).

5. Benchmark Performance (`100% Pass Rate`)

Evaluated on an 8GB RAM CPU environment (6 CPU threads, AVX2):

Domain	Task	Accuracy	Throughput	Verdict
Python	Optimal $O(n)$ Two Sum (3 live unit tests)	100%	16.2 tok/s	PASS
SQL	Multi-table `JOIN`, `GROUP BY`, `HAVING`, `ORDER BY` (SQLite)	100%	12.1 tok/s	PASS
Git	Interactive commit squashing (`git rebase -i HEAD~3`)	100%	17.2 tok/s	PASS
Terminal	Large file search (`find +100M`), RAM (`free -h`), port PID (`lsof`)	100%	16.6 tok/s	PASS
Refactoring	Deeply nested conditionals $\rightarrow$ flat Guard Clauses	100%	15.0 tok/s	PASS
Architecture	100k events/sec Notification System (Kafka, Redis, FCM/WS, DLQ)	100%	16.5 tok/s	PASS
Tool Calling	Structured XML `<tool_call>` emission & schema check	100%	7.8 tok/s	PASS
SWE Tasks	Autonomous file creation (`write_file`) + live subprocess unit test	100%	12.2 tok/s	PASS

Downloads last month: -

GGUF

Model size

3B params

Architecture

granite

Hardware compatibility

4-bit

Model tree for SiddardhaShayini/TAEL_Agent

Base model

ibm-granite/granite-3.1-2b-base

Finetuned

ibm-granite/granite-3.1-2b-instruct

Quantized

(32)

this model

SiddardhaShayini
/

TAEL_Agent

TAEL-Agent (`v2.0.0-rc1`) — Granite 3.1 2B Instruct GGUF (`Q4_K_M`)

1. Repository Files (`SiddardhaShayini/TAEL_Agent`)

2. Quickstart with Ollama

3. Quickstart with `llama-cpp-python`

4. 3-Phase Progressive Curriculum Fine-Tuning (`29,900` Samples)

5. Benchmark Performance (`100% Pass Rate`)

Model tree for SiddardhaShayini/TAEL_Agent

Datasets used to train SiddardhaShayini/TAEL_Agent

TAEL-Agent (v2.0.0-rc1) — Granite 3.1 2B Instruct GGUF (Q4_K_M)

1. Repository Files (SiddardhaShayini/TAEL_Agent)

2. Quickstart with Ollama

3. Quickstart with llama-cpp-python

4. 3-Phase Progressive Curriculum Fine-Tuning (29,900 Samples)

5. Benchmark Performance (100% Pass Rate)

Model tree for SiddardhaShayini/TAEL_Agent

Datasets used to train SiddardhaShayini/TAEL_Agent

TAEL-Agent (`v2.0.0-rc1`) — Granite 3.1 2B Instruct GGUF (`Q4_K_M`)

1. Repository Files (`SiddardhaShayini/TAEL_Agent`)

3. Quickstart with `llama-cpp-python`

4. 3-Phase Progressive Curriculum Fine-Tuning (`29,900` Samples)

5. Benchmark Performance (`100% Pass Rate`)