Instructions to use rbentaarit/kubelm-qwen3.5-2b-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="rbentaarit/kubelm-qwen3.5-2b-v1",
	filename="kubelm-edge.Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf rbentaarit/kubelm-qwen3.5-2b-v1: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 rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf rbentaarit/kubelm-qwen3.5-2b-v1: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 rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

Use Docker

docker model run hf.co/rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

LM Studio
Jan
Ollama
How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Ollama:
```
ollama run hf.co/rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
```

Unsloth Studio

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for rbentaarit/kubelm-qwen3.5-2b-v1 to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for rbentaarit/kubelm-qwen3.5-2b-v1 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for rbentaarit/kubelm-qwen3.5-2b-v1 to start chatting

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf rbentaarit/kubelm-qwen3.5-2b-v1: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 rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Docker Model Runner:
```
docker model run hf.co/rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M
```

Lemonade

How to use rbentaarit/kubelm-qwen3.5-2b-v1 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull rbentaarit/kubelm-qwen3.5-2b-v1:Q4_K_M

Run and chat with the model

lemonade run user.kubelm-qwen3.5-2b-v1-Q4_K_M

List all available models

lemonade list

rbentaarit commited on 4 days ago

Commit

30e7b44

verified ·

1 Parent(s): cfe1637

Card: canonical naming + note adapter now in-repo under adapter/

Browse files

Files changed (1) hide show

README.md +18 -15

README.md CHANGED Viewed

@@ -15,20 +15,21 @@ tags:
 - gguf
 ---
-# kubelm-edge-v0.3 — Q4_K_M GGUF
 A 2B parameter K8sGPT MCP tool-use specialist, trained with QLoRA on
 Qwen3.5-2B and quantized to Q4_K_M for CPU-only deployment. The
-headline deployable of the [kubelm](https://github.com/rbentaarit/kubelm)
-project — supersedes
-[`kubelm-edge-v0`](https://huggingface.co/rbentaarit/kubelm-edge-v0-GGUF).
 ## TL;DR
 On the 35-scenario v0.3 evaluation library, served via `llama-server`
 at temperature 0:
-| metric | qwen2.5-7b (reference) | kubelm-edge-v0 + corrected prompt | **kubelm-edge-v0.3** |
 |---|---|---|---|
 | `conclusion_rubric_passed` | 28 / 35 | 29 / 35 | **32 / 35** |
 | `reference_calls_passed` | 28 / 35 | 27 / 35 | **32 / 35** |
@@ -50,7 +51,7 @@ ollama 0.23.1's `qwen3next` loader currently rejects this GGUF (see
 ```bash
 # Boot the model (Apple Silicon shown; on Linux drop -ngl or set 0)
 brew install llama.cpp   # or: build from https://github.com/ggml-org/llama.cpp
-huggingface-cli download rbentaarit/kubelm-edge-v0.3-GGUF \
     kubelm-edge.Q4_K_M.gguf --local-dir .
 llama-server \
@@ -83,7 +84,7 @@ Sample chat-completion call with a K8sGPT MCP tool:
 curl -sS http://127.0.0.1:8088/v1/chat/completions \
   -H 'Content-Type: application/json' \
   -d '{
-    "model": "kubelm-edge-v0.3",
     "temperature": 0.0,
     "max_tokens": 2048,
     "chat_template_kwargs": {"enable_thinking": false},
@@ -105,8 +106,8 @@ so the model can call real tools against a real cluster.
 - **Tool-use specialist** for K8sGPT MCP investigations on CPU-only
   hardware (M-series Macs, modest Linux boxes).
-- Drop-in upgrade from `kubelm-edge-v0` for K8sGPT integrations that
-  already speak the OpenAI Chat Completions API.
 - Local component of agentic K8s diagnosis pipelines where the
   destructive-action layer is handled by K8sGPT's operator + Mutation
   CR policy gates (i.e. **the model proposes; the operator gates**).
@@ -136,7 +137,8 @@ so the model can call real tools against a real cluster.
   [dataset card](https://huggingface.co/datasets/rbentaarit/kubelm-seed-v0)
   "v0.2 corpus" section for the full provenance.
 - **Method:** QLoRA, rank 32 / alpha 64, target modules
-  `q_proj k_proj v_proj o_proj gate_proj up_proj down_proj`.
 - **Schedule:** 1 epoch, batch 8 × grad-accum 2, lr 2e-4 cosine,
   warmup 3%, max_seq_length 16384, seed 42. Train loss bottomed at
   0.14–0.17 (no overfit; v0.2 on Qwen 2.5 1.5B bottomed at 0.024 and
@@ -190,8 +192,9 @@ Full bench summary (rows for all four columns, every scenario):
   Qwen 3.5 loader stabilizes.
 - **CPU latency on weak hardware.** Per-turn latency on M1 Max with
   Metal offload is ~1.5–2 s; on a 2-core / 2 GB edge box without
-  hardware acceleration, expect single-digit seconds per turn. For
-  per-step latency budgets &lt; 1 s, see `kubelm-edge-v0` (1.5B Qwen 2.5).
 - **No native tool-call format other than OpenAI Chat Completions.**
   Anthropic-style tool-use, Cohere-style, and custom XML formats are
   not trained. Use a translation layer.
@@ -205,11 +208,11 @@ model is Qwen 3.5 2B (Apache 2.0). The training corpus is
 ## Citation
 ```
-@misc{kubelm_edge_v03,
-  title  = {kubelm-edge-v0.3},
   author = {Ramzi Ben Taarit and contributors},
   year   = {2026},
-  url    = {https://huggingface.co/rbentaarit/kubelm-edge-v0.3-GGUF},
   note   = {QLoRA on Qwen3.5-2B; trained against K8sGPT v0.4.32 MCP trajectories}
 }
 ```

 - gguf
 ---
+# kubelm-qwen3.5-2b-v1 — Q4_K_M GGUF
 A 2B parameter K8sGPT MCP tool-use specialist, trained with QLoRA on
 Qwen3.5-2B and quantized to Q4_K_M for CPU-only deployment. The
+headline deployable (**edge+** tier) of the
+[kubelm](https://github.com/rbentaarit/kubelm) project — supersedes the
+edge tier
+[`kubelm-qwen2.5-1.5b-v1`](https://huggingface.co/rbentaarit/kubelm-qwen2.5-1.5b-v1).
 ## TL;DR
 On the 35-scenario v0.3 evaluation library, served via `llama-server`
 at temperature 0:
+| metric | qwen2.5-7b (reference) | kubelm-qwen2.5-1.5b-v1 (edge) | **kubelm-qwen3.5-2b-v1** |
 |---|---|---|---|
 | `conclusion_rubric_passed` | 28 / 35 | 29 / 35 | **32 / 35** |
 | `reference_calls_passed` | 28 / 35 | 27 / 35 | **32 / 35** |
 ```bash
 # Boot the model (Apple Silicon shown; on Linux drop -ngl or set 0)
 brew install llama.cpp   # or: build from https://github.com/ggml-org/llama.cpp
+huggingface-cli download rbentaarit/kubelm-qwen3.5-2b-v1 \
     kubelm-edge.Q4_K_M.gguf --local-dir .
 llama-server \
 curl -sS http://127.0.0.1:8088/v1/chat/completions \
   -H 'Content-Type: application/json' \
   -d '{
+    "model": "kubelm-qwen3.5-2b",
     "temperature": 0.0,
     "max_tokens": 2048,
     "chat_template_kwargs": {"enable_thinking": false},
 - **Tool-use specialist** for K8sGPT MCP investigations on CPU-only
   hardware (M-series Macs, modest Linux boxes).
+- Drop-in upgrade from `kubelm-qwen2.5-1.5b-v1` for K8sGPT integrations
+  that already speak the OpenAI Chat Completions API.
 - Local component of agentic K8s diagnosis pipelines where the
   destructive-action layer is handled by K8sGPT's operator + Mutation
   CR policy gates (i.e. **the model proposes; the operator gates**).
   [dataset card](https://huggingface.co/datasets/rbentaarit/kubelm-seed-v0)
   "v0.2 corpus" section for the full provenance.
 - **Method:** QLoRA, rank 32 / alpha 64, target modules
+  `q_proj k_proj v_proj o_proj gate_proj up_proj down_proj`. LoRA
+  adapter included in this repo under `adapter/`.
 - **Schedule:** 1 epoch, batch 8 × grad-accum 2, lr 2e-4 cosine,
   warmup 3%, max_seq_length 16384, seed 42. Train loss bottomed at
   0.14–0.17 (no overfit; v0.2 on Qwen 2.5 1.5B bottomed at 0.024 and
   Qwen 3.5 loader stabilizes.
 - **CPU latency on weak hardware.** Per-turn latency on M1 Max with
   Metal offload is ~1.5–2 s; on a 2-core / 2 GB edge box without
+  hardware acceleration, expect single-digit seconds per turn. For the
+  lowest per-step latency and smallest footprint, see the ultra-edge
+  `kubelm-qwen3.5-0.8b-v1`.
 - **No native tool-call format other than OpenAI Chat Completions.**
   Anthropic-style tool-use, Cohere-style, and custom XML formats are
   not trained. Use a translation layer.
 ## Citation
 ```
+@misc{kubelm_qwen35_2b_v1,
+  title  = {kubelm-qwen3.5-2b-v1},
   author = {Ramzi Ben Taarit and contributors},
   year   = {2026},
+  url    = {https://huggingface.co/rbentaarit/kubelm-qwen3.5-2b-v1},
   note   = {QLoRA on Qwen3.5-2B; trained against K8sGPT v0.4.32 MCP trajectories}
 }
 ```