Instructions to use kortexa-ai/shingi-27b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use kortexa-ai/shingi-27b 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 kortexa-ai/shingi-27b
# Run inference directly in the terminal:
llama cli -hf kortexa-ai/shingi-27b

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf kortexa-ai/shingi-27b
# Run inference directly in the terminal:
llama cli -hf kortexa-ai/shingi-27b

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf kortexa-ai/shingi-27b
# Run inference directly in the terminal:
./build/bin/llama cli -hf kortexa-ai/shingi-27b

Use Docker

docker model run hf.co/kortexa-ai/shingi-27b

Llama new
LM Studio
Jan
Ollama
How to use kortexa-ai/shingi-27b with Ollama:
```
ollama run hf.co/kortexa-ai/shingi-27b
```
Unsloth Desktop

How to use kortexa-ai/shingi-27b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf kortexa-ai/shingi-27b

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": "kortexa-ai/shingi-27b"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use kortexa-ai/shingi-27b with Docker Model Runner:
```
docker model run hf.co/kortexa-ai/shingi-27b
```

Lemonade

How to use kortexa-ai/shingi-27b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull kortexa-ai/shingi-27b

Run and chat with the model

lemonade run user.shingi-27b-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use kortexa-ai/shingi-27b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf kortexa-ai/shingi-27b

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 kortexa-ai/shingi-27b

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use kortexa-ai/shingi-27b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf kortexa-ai/shingi-27b

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 "kortexa-ai/shingi-27b" \
  --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"

francip commited on 25 days ago

Commit

d02406f

verified ·

1 Parent(s): 94b9706

Shingi 27B v3.2 weights: stage 2 on data v3.2 (scales-04) and its calibration

Browse files

Files changed (2) hide show

calibration.json +22 -18
shingi-27b.gguf +1 -1

calibration.json CHANGED Viewed

@@ -1,12 +1,12 @@
 {
   "parameters": {
     "temperature": 1.0,
-    "noul_temperature": 1.25,
     "noul_bias": 0.0
   },
-  "fit_records": 1276,
-  "choice_score_records": 1056,
-  "noul_records": 220,
   "grid": {
     "temperatures": [
       0.5,
@@ -25,24 +25,20 @@
     ]
   },
   "before": {
-    "choice_score_nll": 0.6340668104525349,
-    "noul_nll": 0.3180781955467112
   },
   "after": {
-    "choice_score_nll": 0.6340668104525349,
-    "noul_nll": 0.3072396297437681
   },
   "limitations": "Global temperatures only; no weight changes. The yes/no bias is not fitted because the split's yes/no pool is small and its gold-yes share reflects a few sources.",
   "objective": "hard-label NLL on the frozen data v3 calibration split only; temperatures only, yes/no bias fixed at 0",
   "noul_pool": {
-    "records": 220,
-    "gold_yes": 110,
     "gold_yes_share": 0.5,
     "by_source": {
-      "civil_comments": {
-        "records": 40,
-        "gold_yes": 20
-      },
       "gsm8k_judge": {
         "records": 80,
         "gold_yes": 40
@@ -62,16 +58,24 @@
       "synth_state": {
         "records": 14,
         "gold_yes": 7
       }
     }
   },
   "provenance": {
-    "source_revision": "18e4235aae7a8d449d1eb9098bbbe10305790574",
-    "model_sha256": "277d970372fac37185bb621341d99bec3c7f1d86c6657d27ca18a6dcf930bd03",
     "adapter_sha256": null,
     "executable_sha256": "dd71ef5ac08ac18e2028db3dd84dcca7864f12b904fb9a76379c12c0f034c2fb",
-    "data_manifest_sha256": "1e3ebc747608e7010e6f83e9ae203e41892851ab174391c6fcffa5efa9d8d4af",
     "choice_order": "canonical",
-    "predictions_sha256": "9a5d58973b1f04985020b54d5ac0921a7f27ae5dad6df931e81338d48f040aa0"
   }
 }

 {
   "parameters": {
     "temperature": 1.0,
+    "noul_temperature": 1.0,
     "noul_bias": 0.0
   },
+  "fit_records": 1696,
+  "choice_score_records": 1356,
+  "noul_records": 340,
   "grid": {
     "temperatures": [
       0.5,
     ]
   },
   "before": {
+    "choice_score_nll": 0.6331164448410751,
+    "noul_nll": 0.20858523761769068
   },
   "after": {
+    "choice_score_nll": 0.6331164448410751,
+    "noul_nll": 0.20858523761769068
   },
   "limitations": "Global temperatures only; no weight changes. The yes/no bias is not fitted because the split's yes/no pool is small and its gold-yes share reflects a few sources.",
   "objective": "hard-label NLL on the frozen data v3 calibration split only; temperatures only, yes/no bias fixed at 0",
   "noul_pool": {
+    "records": 340,
+    "gold_yes": 170,
     "gold_yes_share": 0.5,
     "by_source": {
       "gsm8k_judge": {
         "records": 80,
         "gold_yes": 40
       "synth_state": {
         "records": 14,
         "gold_yes": 7
+      },
+      "tabfact": {
+        "records": 80,
+        "gold_yes": 40
+      },
+      "tat_qa_judge": {
+        "records": 80,
+        "gold_yes": 40
       }
     }
   },
   "provenance": {
+    "source_revision": "a0e9b38bbac543ddef50a3aba4c1d799f78a679e",
+    "model_sha256": "c62ae5b61e458aa70aa2f51ebbd193d50e00d170cb318c60341bb68c6471242c",
     "adapter_sha256": null,
     "executable_sha256": "dd71ef5ac08ac18e2028db3dd84dcca7864f12b904fb9a76379c12c0f034c2fb",
+    "data_manifest_sha256": "c6eeb7ddf21a94335d6ea8c5b9ea77cc3f1f12db41220c04924d19206542314e",
     "choice_order": "canonical",
+    "predictions_sha256": "52c45b33e8a75548de7e65b4b99d58bff1935487153701e30f36101ff852a403"
   }
 }

shingi-27b.gguf CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:277d970372fac37185bb621341d99bec3c7f1d86c6657d27ca18a6dcf930bd03
 size 7206168928

 version https://git-lfs.github.com/spec/v1
+oid sha256:c62ae5b61e458aa70aa2f51ebbd193d50e00d170cb318c60341bb68c6471242c
 size 7206168928