tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-15 04:33:06 -04:00
Code Issues Packages Projects Releases Wiki Activity

Add LLaDA 8b Diffusion model (#14771)

Browse Source 
* Add support for Llada-8b: diffusion model

* Add README

* Fix README and convert_hf_to_gguf

* convert_hf_to_gguf.py: address review comments

* Make everything in a single example

* Remove model-specific sampling

* Remove unused argmax

* Remove braced initializers, improve README.md a bit

* Add diffusion specific gguf params in set_vocab, remove setting rope_theta and rms_norm_eps

* Remove adding the mask token

* Move add_add_bos_token to set_vocab

* use add_bool in gguf_writer.py
This commit is contained in:
Aman Gupta
2025-07-31 19:49:09 +08:00
committed by GitHub
parent 11490b3672
commit 8a4a856277
12 changed files with 931 additions and 385 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
common/common.h
View File

@@ -220,11 +220,17 @@ struct common_params_vocoder {
};
struct common_params_diffusion {
int32_t steps = 64; // number of diffusion steps
float eps = 1e-3f; // epsilon for timesteps
int32_t algorithm = 0; // diffusion algorithm (0=ORIGIN, 1=MASKGIT_PLUS, 2=TOPK_MARGIN, 3=ENTROPY)
float alg_temp = 0.0f; // algorithm temperature
bool visual_mode = false; // show progressive diffusion on screen
int32_t steps = 128;
bool visual_mode = false;
float eps = 0; // epsilon for timesteps
int32_t block_length = 32; // block length for generation
int32_t algorithm = 4; // default algorithm: low-confidence
float alg_temp = 0.0f; // algorithm temperature
float cfg_scale = 0; // classifier-free guidance scale
bool add_gumbel_noise = false; // add gumbel noise to the logits if temp > 0.0
};
enum common_reasoning_format {