Skip to content

Commit 67d5ecf

Browse files
committed
fix(core): add GLM-4.7-Flash-8bit template with glm47_flash parser
- Update version to 0.1.11 in README and script - Add GLM-4.7-Flash-8bit option to template model selection menu - Configure reasoning-parser as glm47_flash for GLM-4.7-Flash-8bit - Update usage instructions with new template model information - Reorder template choices for better organization
1 parent b56f167 commit 67d5ecf

File tree

2 files changed

+53
-21
lines changed

2 files changed

+53
-21
lines changed

README.md

Lines changed: 39 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@ MLX Hugging Face Manager
99
by Laurent Marques
1010
```
1111

12-
**Version: 0.1.10**
12+
**Version: 0.1.11**
1313

1414
A terminal-based LLM model manager for Apple Silicon Macs. Browse, install, and run MLX-optimized models from Hugging Face with an intuitive TUI interface.
1515

@@ -89,15 +89,44 @@ brew install huggingface-cli
8989
- Python 3.11 or 3.12 (Python 3.13+ is not yet supported)
9090
- Use pyenv to manage Python versions if needed
9191

92-
#### Installation Method 1: Using pipx (Recommended)
92+
#### Installation Method 1: Using uv (Recommended)
9393

9494
```bash
95-
# Install mlx-openai-server
96-
pipx install mlx-openai-server
95+
```bash
9796
98-
# Update mlx-lm to latest version for newest model support (e.g., iquestcoder)
99-
pipx runpip mlx-openai-server uninstall mlx-lm -y
100-
pipx runpip mlx-openai-server install git+https://github.com/ml-explore/mlx-lm.git
97+
98+
# Set your preferred Python version
99+
pyenv global 3.12.8
100+
101+
cd ~/Scripts/mlx-tools
102+
103+
# Clone the mlx-openai-server repository
104+
git clone https://github.com/cubist38/mlx-openai-server.git
105+
cd mlx-openai-server
106+
107+
# Create virtual environment (will use your pyenv Python 3.12.8)
108+
uv venv
109+
110+
# Activate the environment
111+
source .venv/bin/activate
112+
113+
# Install in development mode
114+
uv pip install -e .
115+
116+
# Install latest mlx-lm from GitHub if needed to update to last mlx-lm
117+
# uv pip install git+https://github.com/ml-explore/mlx-lm.git
118+
119+
```
120+
Then
121+
> add in your zshrc
122+
```bash
123+
mlx-openai-server() {
124+
local original_dir="$PWD"
125+
cd ~/Scripts/mlx-tools/mlx-openai-server
126+
source .venv/bin/activate
127+
command mlx-openai-server "$@"
128+
cd "$original_dir"
129+
}
101130
```
102131

103132
**Verify installation:**
@@ -349,6 +378,9 @@ $MODEL_DIR/ # Selected path (External/Local/Legacy)
349378
- Added template models documentation with usage instructions
350379
- Improved main menu organization and discoverability
351380
381+
**Fix (v0.1.11):**
382+
- Added **glm47_flash** to --reasoning-parser param from GLM-4.7-Flash-8bit template
383+
352384
## Keyboard Shortcuts
353385
354386
| Key | Action |

src/efx-face-manager.sh

Lines changed: 14 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@
77
# Uses gum for interactive selection
88
# https://github.com/charmbracelet/gum
99

10-
VERSION="0.1.10"
10+
VERSION="0.1.11"
1111

1212
clear
1313

@@ -1276,12 +1276,23 @@ while true; do
12761276
# Fast launch predefined template models
12771277
template_choice=$(gum choose \
12781278
--header "Select a template model to run" \
1279-
"Qwen3-Coder-30B-A3B-Instruct-8bit" \
1280-
"NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-8Bit" \
12811279
"GLM-4.7-Flash-8bit" \
1280+
"NVIDIA-Nemotron-3-Nano-30B-A3B-MLX-8Bit" \
1281+
"Qwen3-Coder-30B-A3B-Instruct-8bit" \
12821282
"✖ Back")
12831283

12841284
case "$template_choice" in
1285+
"GLM-4.7-Flash-8bit")
1286+
CMD_ARGS=("--model-path" "$MODEL_DIR/GLM-4.7-Flash-8bit")
1287+
CMD_ARGS+=("--model-type" "lm")
1288+
CMD_ARGS+=("--reasoning-parser" "glm47_flash")
1289+
CMD_ARGS+=("--tool-call-parser" "glm4_moe")
1290+
CMD_ARGS+=("--message-converter" "glm4_moe")
1291+
CMD_ARGS+=("--debug")
1292+
CMD_ARGS+=("--port" "8000")
1293+
CMD_ARGS+=("--host" "0.0.0.0")
1294+
confirm_and_launch
1295+
;;
12851296
"Qwen3-Coder-30B-A3B-Instruct-8bit")
12861297
CMD_ARGS=("--model-path" "$MODEL_DIR/Qwen3-Coder-30B-A3B-Instruct-8bit")
12871298
CMD_ARGS+=("--model-type" "lm")
@@ -1301,17 +1312,6 @@ while true; do
13011312
CMD_ARGS+=("--trust-remote-code")
13021313
confirm_and_launch
13031314
;;
1304-
"GLM-4.7-Flash-8bit")
1305-
CMD_ARGS=("--model-path" "$MODEL_DIR/GLM-4.7-Flash-8bit")
1306-
CMD_ARGS+=("--model-type" "lm")
1307-
CMD_ARGS+=("--reasoning-parser" "glm4_moe") # glm47_flash
1308-
CMD_ARGS+=("--tool-call-parser" "glm4_moe")
1309-
CMD_ARGS+=("--message-converter" "glm4_moe")
1310-
CMD_ARGS+=("--debug")
1311-
CMD_ARGS+=("--port" "8000")
1312-
CMD_ARGS+=("--host" "0.0.0.0")
1313-
confirm_and_launch
1314-
;;
13151315
esac
13161316
;;
13171317

0 commit comments

Comments
 (0)