Upload README.md with huggingface_hub

README.md

@@ -79,6 +79,18 @@ Recommended sampling (from the official model card):
 `top_p = 1`, `temperature = 0`. For coding tasks the authors suggest
 `alg = entropy, num_small_blocks = 32 (→ --diffusion-small-block 1), top_p = 0.8, temperature = 1`.
 
+**Fast mode** (higher throughput): reveal every masked token whose confidence exceeds a
+threshold in a single denoising step, with one small block per block:
+
+```bash
+./build/bin/llama-diffusion-cli \
+  -m pangu-diffusion-Q4_K_M.gguf \
+  -p "[unused9]系统：[unused10][unused9]用户：<question>[unused10][unused9]助手：" \
+  -ub 256 --diffusion-block-length 32 --diffusion-small-block 32 \
+  --diffusion-confidence-threshold 0.9 --diffusion-algorithm 4 \
+  --temp 0 -t 16 --seed 42
+```
+
 `--diffusion-algorithm` values: `0` origin, `1` entropy, `2` margin, `3` random, `4` confidence.
 `-ub N` is the maximum sequence length; the tail that does not fill a whole block is dropped.
 
@@ -139,16 +151,18 @@ experience while using it:
 |---|---|---|
 | **Input** — waiting for the prompt to be read before output starts | both | **≈ 78 t/s** short prompt; **≈ 132 t/s** for longer prompts (`llama-bench` pp512) |
 | **Output** — text generated (autoregressive, streamed live) | `llama-cli` | **11.9 t/s** (llama-bench tg128: 11.1 t/s) |
-| **Output** — text generated (block diffusion, block 32 / small-block 4, `alg=entropy`, `temp=0`) | `llama-diffusion-cli` | **≈ 4.4 t/s** |
+| **Output** — block diffusion (quality: block 32 / small-block 4, `alg=entropy`, `temp=0`) | `llama-diffusion-cli` | **≈ 4.4 t/s** |
+| **Output** — block diffusion (fast: `--diffusion-confidence-threshold 0.9`, small-block 32, `alg=confidence`) | `llama-diffusion-cli` | **≈ 5.2 t/s** |
 
 Notes:
 - Input (prompt processing) is the same causal prefill for both paths; it is what you wait
   for before the answer appears.
 - The AR output speed (≈ 12 t/s) is what `llama-cli` prints as `Generation: X t/s`; the
   diffused text is only flushed at the end of a run (unless `--diffusion-visual`).
-- Block diffusion reveals one token per denoising step, i.e. up to `block_length` (= 32)
-  forward passes per block, so its tokens/s is inherently lower. The authors' faster
-  `alg=confidence_threshold` mode is not implemented here (fixed 1-token steps).
+- Block diffusion normally reveals one token per denoising step, i.e. up to `block_length`
+  (= 32) forward passes per block. With `--diffusion-confidence-threshold T` every masked
+  token whose confidence exceeds `T` is revealed in the same step, which raises throughput
+  (the authors report up to 2.5× vs. autoregressive on Ascend NPUs).
 - The AR figures are the best reference for raw model throughput on this hardware.
 
 ## What the patch does
@@ -191,6 +205,28 @@ tokenizer.ggml.mask_token_id    = 45830
 - Quantization may slightly affect the tricky reasoning cases; use `alg = entropy`,
   `temp = 0` for the most stable results.
 
+## Supported features
+
+Implemented (matching the reference `inference/generation_utils.py`):
+
+| Feature | Status |
+|---|---|
+| Context-causal block diffusion — full attention inside a block, causal attention over the preceding context | ✓ |
+| Block-by-block decoding (block length 32) | ✓ `--diffusion-block-length` |
+| Committed-block KV cache with a causal mask; first token of the next block decoded from the last position | ✓ |
+| Small blocks inside a block (`num_small_blocks`, optimum 4 or 8) | ✓ `--diffusion-small-block` |
+| Variable-length inference / flexible context | ✓ |
+| Logit shift (`diffusion.shift_logits`) | ✓ |
+| Dedicated mask token (`tokenizer.ggml.mask_token_id = 45830`) | ✓ |
+| Reveal order: `confidence`, `entropy`, `topk_margin` | ✓ `--diffusion-algorithm 4/1/2` |
+| `confidence_threshold` fast mode (multi-token reveal per step) | ✓ `--diffusion-confidence-threshold` |
+| EOS early stop | ✓ |
+| Autoregressive decoding | ✓ `llama-cli` |
+
+Not implemented:
+- Batched (multi-sequence) block diffusion — the example CLI runs a single sequence.
+- Vendor NPU kernels — this uses whatever backend llama.cpp was built with (CPU here).
+
 ## Original model
 
 - Weights: [`openpangu/openPangu-R-7B-Diffusion`](https://huggingface.co/openpangu/openPangu-R-7B-Diffusion)