GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
pinkelephantlimited
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
pinkelephantlimited
Model Tree
Input Modalities
Output Modalities
Supported Functionality
| Parameter | Value |
|---|---|
| Architecture | LLaMA (decoder-only) |
| Parameters | 32,514,560 |
| Hidden size | 512 |
| Intermediate size | 1,792 |
| Num layers | 8 |
| Num attention heads | 8 |
| Num KV heads | 8 |
| Max sequence length | 2,048 |
| Vocabulary size | 4,096 |
| Position encoding | RoPE (θ=10000.0) |
| Activation | SwiGLU |
| Normalization | Pre-RMSNorm |
| Weight tying | Yes (embedding ↔ lm_head) |
| Precision | fp32 |
| Source | Type | Size |
|---|---|---|
| FineWeb-Edu (sample-10BT) | Educational web text | 150,000 examples |
| Parameter | Value |
|---|---|
| Hardware | 96GB GPU |
| Optimizer | AdamW (8-bit) |
| Learning rate | 2e-4 |
| Batch size | 32 |
| Gradient accumulation | 4 |
| Effective batch size | 128 |
| Warmup steps | 500 |
| Weight decay | 0.01 |
| Training time | ~30 minutes |
python
from transformers import pipelinepipe = pipeline("text-generation", model="pinkelephantlimited/pink-elephant-33m")output = pipe("The water cycle consists of", max_new_tokens=80)[0]["generated_text"]print(output)
This model was trained entirely from scratch on permissively licensed data. No fine-tuning, no transfer learning — 100% original weights.
MIT
Deploy on school tablets, offline kiosks, or mobile apps:
python
from transformers import pipelinepipe = pipeline("text-generation", model="pinkelephantlimited/pink-elephant-33m")# Homework help — explain conceptspipe("The water cycle consists of three main stages:", max_new_tokens=80)pipe("Photosynthesis is the process by which", max_new_tokens=80)
python
# Draft sentences for journaling apps or note-taking toolspipe("In conclusion, the key factors that led to", max_new_tokens=60)
python
# Customer service triage bot — runs on a standard VPSpipe("Thank you for contacting support. To help you better,", max_new_tokens=80)
Why 33M? Runs on CPU at ~50ms/token, fits in 200MB RAM. Perfect for serverless/edge deployments where GPU is unavailable.