Downloads · 30 days
0
contact-center-coaching-mistral-7b-mlx
chendren/contact-center-coaching-mistral-7b-mlx
contact-center-coaching-mistral-7b-mlx is a text generation model from chendren. Use it when you need the model to write or continue text. It is set up for mlx. The card lists the license as apache-2.0.
LoRA on mlx-community/Mistral-7B-Instruct-v0.3 for journey-tagged coaching JSON (onboarding, escalation, retention, sales) with temporal context.
Access
Public
Updated Oct 23, 2026
Repo size
42 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors83.9 MB · 100%
At a glance
- Task
- Text Generation
- Library
- mlx
- License
- apache-2.0
- Access
- Public
- Created
- Jan 2, 2026
- Updated
- Oct 23, 2026
- SHA
- 790ea779
Try a prompt
Base models
Model card
From the Hugging Face model README
Contact Center Coaching, Mistral 7B LoRA (MLX)
LoRA on mlx-community/Mistral-7B-Instruct-v0.3 for journey-tagged coaching JSON (onboarding, escalation, retention, sales) with temporal context.
Research. Training data is synthetic (InceptionLabs Mercury). The synthetic set is not on the Hub. Test perplexity 1.35 is not a coaching-quality metric; it is consistent with copying short JSON targets. No held-out exact-match or human rater score is published.
For process-control gold see Immortalis/cx-comply.
Load
The release file is adapters.safetensors (copy of the 1,200-step adapter).
from mlx_lm import load, generate
model, tokenizer = load(
"mlx-community/Mistral-7B-Instruct-v0.3",
adapter_path="chendren/contact-center-coaching-mistral-7b-mlx",
)
Training
| Parameter | Value |
|---|---|
| Method | LoRA, 0.145% trainable (10.5M / 7.2B) |
| Iterations | 1,200 |
| Batch | 4 |
| LR | 1e-5 |
| Hardware | M4 Max, MLX |
| Validation loss | 0.268 |
| Test perplexity | 1.35 (likely overfit to synthetic JSON) |
Limits
- Synthetic conversations only
- Do not use outputs for agent performance management
- Prefer exact-match on a held-out schema before any desk trial
- Task
- Text Generation
- Library
- mlx
- License
- apache-2.0
- Languages
- en
- Created
- Jan 2, 2026
- Updated
- Oct 23, 2026