GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
Crusadersk
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
Crusadersk
Model Tree
Input Modalities
Output Modalities
Supported Functionality
Mid-scale MHA model for scaling-regime validation.
These checkpoints are not general-purpose language models. They are deliberately sized scaling-study artifacts designed to isolate the effect of model depth vs width on GPU inference latency. The key finding: in the small-model GPU regime, layer depth (not parameter count) dominates latency, producing inversions where a 5M-parameter model can be 3.6x slower than a 25M-parameter model.
Used in: TR117, TR126, TR147
| TR | Role |
|---|---|
| TR117 | Original cross-backend benchmark matrix (7 backends, 4 model groups) |
| TR126 | Linux/Triton compiler validation with phase-separated measurement |
| TR147 | Second-regime portability validation on RTX 6000 Ada |
The GPT-2 family (25M, 50M, 100M) uses a 2x3 factorial design:
| Model | n_embd | n_layer | n_inner | Params | Design role |
|---|---|---|---|---|---|
| gpt2-25m | 384 | 3 | 1,536 | 25M | Shallow, narrow |
| gpt2-50m | 512 | 8 | 2,048 | 50M | Deep, medium width |
| gpt2-100m | 768 | 8 | 3,072 | 100M | Deep, wide |
All models use 2 attention heads (MHA, not GQA) to isolate architecture effects from attention-group structure. Dropout is set to 0.0 for deterministic inference measurement.
python
from transformers import AutoModelForCausalLM, AutoTokenizermodel = AutoModelForCausalLM.from_pretrained("Crusadersk/gpt2-50m")tokenizer = AutoTokenizer.from_pretrained("Crusadersk/gpt2-50m")inputs = tokenizer("Hello", return_tensors="pt")outputs = model.generate(**inputs, max_new_tokens=32, do_sample=False)print(tokenizer.decode(outputs[0], skip_special_tokens=True))
| Framework | Supported |
|---|---|
| Transformers | Yes |
| torch.compile (Inductor) | Yes |
| Ollama | No (not GGUF format) |
| vLLM | Yes |
bibtex
@misc{banterhearts2026gpt250m,title = {Custom GPT-2 Scaling Checkpoint (50M) for Inference Benchmarking Research},author = {Kadadekar, Sahil},year = {2026},url = {https://huggingface.co/Crusadersk/gpt2-50m},note = {Part of the Banterhearts research program. NeurIPS 2026 submission.}}
This work is part of the Chimera/Banterhearts technical-report program on deployment-time LLM behavior, quantization, refusal robustness, batching effects, and inference-stack reliability. Canonical public archive: Chimeraforge Reports; source context: github.com/Sahil170595/Banterhearts.