Model: Qwen3 Next #16095

pwilkin · 2025-09-18T19:45:57Z

EDIT: README FIRST
This is an implementation of a new type of attention gating in GGML.
Therefore, this implementation will be focused on CORRECTNESS ONLY.
Speed tuning and support for more architectures will come in future PRs.
Please do not spam this threads with reports about performance, especially on backend architectures (CUDA, Vulkan).

CURRENT STATE: core is done

===
It's been a real learning experience, not gonna lie, but if someone with hybrid model implementation experience (@gabe-l-hart ?) has some quick tips, I'd be grateful.

Resolves #15940

gabe-l-hart · 2025-09-18T19:55:31Z

I'll try to get into it in more detail soon, but here are a few general thoughts after quickly skimming the PR:

The structure of what you've got smells correct, so it's likely close, but missing something small yet critical
A full repro with the error it's raising would definitely help debug
My debugging process for this would be:
1. Make sure tokenization is solid (print statements as necessary to compare tokens before input)
2. Use llama-eval-callback to dump tensors for a single prefill step
3. Run an identical single prefill with the reference impl (transformers or otherwise), and inject prints as needed to dump tensors along the way
4. Visually comb through them (particularly the sum at each point) to see where things start diverging significantly

bugparty · 2025-09-18T20:47:56Z

It's been a real learning experience, not gonna lie, but if someone with hybrid model implementation experience (@gabe-l-hart ?) has some quick tips, I'd be grateful.

Currently at the stage of "graph builds, but first decode complains about wrong memory model", probably not building the inputs correctly.

Resolves #15940

interesting, maybe we can learn together

ggml/src/ggml.c

ggml/src/ggml-cpu/ggml-cpu.c

pwilkin · 2025-09-19T12:53:24Z

A full repro with the error it's raising would definitely help debug

Running llama-cli -m reference/qwen3_next_500m/Qwen3_Next_500M-8x417M-BF16.gguf -ngl 999 -p "Who are " yields this weird memory error:

#0  __syscall_cancel_arch () at ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S:56
56      in ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S
#1  0x000070552b29eb63 in __internal_syscall_cancel (a1=<optimized out>, a2=<optimized out>, a3=<optimized out>, a4=<optimized out>, a5=0, a6=0, nr=61) at ./nptl/cancellation.c:49
warning: 49     ./nptl/cancellation.c: No such file or directory
#2  __syscall_cancel (a1=<optimized out>, a2=<optimized out>, a3=<optimized out>, a4=<optimized out>, a5=a5@entry=0, a6=a6@entry=0, nr=61) at ./nptl/cancellation.c:75
75      in ./nptl/cancellation.c
#3  0x000070552b31afdf in __GI___wait4 (pid=<optimized out>, stat_loc=<optimized out>, options=<optimized out>, usage=<optimized out>) at ../sysdeps/unix/sysv/linux/wait4.c:30
warning: 30     ../sysdeps/unix/sysv/linux/wait4.c: No such file or directory
#4  0x000070552bb45c31 in ggml_print_backtrace () at /devel/tools/llama.cpp/ggml/src/ggml.c:196
warning: Source file is more recent than executable.
196             waitpid(child_pid, NULL, 0);
#5  0x000070552bb45de5 in ggml_abort (file=0x70552bbcdac8 "/devel/tools/llama.cpp/ggml/src/ggml-backend.cpp", line=189, fmt=0x70552bbcd8af "GGML_ASSERT(%s) failed") at /devel/tools/llama.cpp/ggml/src/ggml.c:230
230             ggml_print_backtrace();
#6  0x000070552bb6091e in ggml_backend_buffer_get_type (buffer=0x0) at /devel/tools/llama.cpp/ggml/src/ggml-backend.cpp:189
189         GGML_ASSERT(buffer);
#7  0x000070552bb6080e in ggml_backend_buffer_is_host (buffer=0x0) at /devel/tools/llama.cpp/ggml/src/ggml-backend.cpp:170
170         return ggml_backend_buft_is_host(ggml_backend_buffer_get_type(buffer));
#8  0x000070552c07a114 in llm_graph_input_rs::set_input (this=0x5f11bdf6aea0, ubatch=0x5f11be011300) at /devel/tools/llama.cpp/src/llama-graph.cpp:241
241             GGML_ASSERT(ggml_backend_buffer_is_host(s_copy->buffer));
#9  0x000070552c07b03c in llm_graph_input_mem_hybrid::set_input (this=0x5f11bdf6aee0, ubatch=0x5f11be011300) at /devel/tools/llama.cpp/src/llama-graph.cpp:437
437         inp_rs->set_input(ubatch);
#10 0x000070552c07b549 in llm_graph_result::set_inputs (this=0x5f11be01ddf0, ubatch=0x5f11be011300) at /devel/tools/llama.cpp/src/llama-graph.cpp:480
480             input->set_input(ubatch);
#11 0x000070552c01ddb3 in llama_context::process_ubatch (this=0x5f11c05b5b50, ubatch=..., gtype=LLM_GRAPH_TYPE_DECODER, mctx=0x5f11be00ff00, ret=@0x7fff74d22ea4: 538976288) at /devel/tools/llama.cpp/src/llama-context.cpp:779
779             res->set_inputs(&ubatch);
#12 0x000070552c01f367 in llama_context::decode (this=0x5f11c05b5b50, batch_inp=...) at /devel/tools/llama.cpp/src/llama-context.cpp:1088
1088            const auto * res = process_ubatch(ubatch, LLM_GRAPH_TYPE_DECODER, mctx.get(), status);
#13 0x000070552c025e49 in llama_decode (ctx=0x5f11c05b5b50, batch=...) at /devel/tools/llama.cpp/src/llama-context.cpp:2726
2726        const int ret = ctx->decode(batch);
#14 0x00005f11a2021559 in common_init_from_params (params=...) at /devel/tools/llama.cpp/common/common.cpp:1066
1066                llama_decode(lctx, llama_batch_get_one(tmp.data(), std::min(tmp.size(), (size_t) params.n_batch)));
#15 0x00005f11a1e4a3c0 in main (argc=7, argv=0x7fff74d25968) at /devel/tools/llama.cpp/tools/main/main.cpp:140
140         common_init_result llama_init = common_init_from_params(params);

I'll try to merge the op into the ggml_delta_net function call as @ngxson suggested.

CISC · 2025-09-19T13:28:30Z

A full repro with the error it's raising would definitely help debug

Running llama-cli -m reference/qwen3_next_500m/Qwen3_Next_500M-8x417M-BF16.gguf -ngl 999 -p "Who are " yields this weird memory error:
...
#6  0x000070552bb6091e in ggml_backend_buffer_get_type (buffer=0x0) at /devel/tools/llama.cpp/ggml/src/ggml-backend.cpp:189
189         GGML_ASSERT(buffer);
#7  0x000070552bb6080e in ggml_backend_buffer_is_host (buffer=0x0) at /devel/tools/llama.cpp/ggml/src/ggml-backend.cpp:170
170         return ggml_backend_buft_is_host(ggml_backend_buffer_get_type(buffer));
...

The backend buffer is NULL.

ngxson · 2025-09-19T13:40:52Z

#9  0x000070552c07b03c in llm_graph_input_mem_hybrid::set_input (this=0x5f11bdf6aee0, ubatch=0x5f11be011300) at /devel/tools/llama.cpp/src/llama-graph.cpp:437
437         inp_rs->set_input(ubatch);

The model doesn't seem to have any recurrence layers. This makes the set input fails due to input node not being present in cgraph.

I'll try to merge the op into the ggml_delta_net function call as @ngxson suggested.

Hmm I think I said the reverse: not to merge it but make the op simple

I feel like this op can be implemented using other ggml ops like mul, mul_mat, sum. Which part of the calculation do you think that can't be constructed using existing ops?

This is the more important question: should we try to implement it using existing ops, or add a new op and spend even more time to optimize it cross all backends?

pwilkin · 2025-09-19T15:10:34Z

Now this is an error I haven't expected to encounter:

GGML_ABORT("not enough space in the context's memory pool");

pwilkin · 2025-09-19T15:18:04Z

The model doesn't seem to have any recurrence layers. This makes the set input fails due to input node not being present in cgraph.

How do I allocate the memory for the linear layers then? I seem to have misunderstood how build_inp_mem_hybrid() works...

yarikdevcom · 2025-09-19T15:34:12Z

@pwilkin any chance to buy you a coffee?(Paterson etc.) so community able to donate for your efforts. Thank you!

pwilkin · 2025-09-19T16:43:14Z

@pwilkin any chance to buy you a coffee?(Paterson etc.) so community able to donate for your efforts. Thank you!

Added a buymeacoffee link to my profile (do consider first funding the Llama.cpp project itself, though!)

ServeurpersoCom · 2025-09-19T16:55:46Z

@pwilkin any chance to buy you a coffee?(Paterson etc.) so community able to donate for your efforts. Thank you!

Added a buymeacoffee link to my profile (do consider first funding the Llama.cpp project itself, though!)

I send a coffee also.

ngxson · 2025-09-20T03:54:48Z

GGML_ABORT("not enough space in the context's memory pool");

Probably there are too many nodes on cgraph, try increasing the limit via llama_context::graph_max_nodes()

src/llama-model.cpp

ngxson · 2025-09-20T05:01:34Z

src/llama-model.cpp

+        Qcur = ggml_reshape_3d(ctx0, ggml_cont(ctx0, Qcur), n_embd_head, hparams.n_head(il), n_tokens);
+        Kcur = ggml_reshape_3d(ctx0, ggml_cont(ctx0, Kcur), n_embd_head, hparams.n_head_kv(il), n_tokens);
+        Vcur = ggml_reshape_3d(ctx0, ggml_cont(ctx0, Vcur), n_embd_head, hparams.n_head_kv(il), n_tokens);


these ggml_cont can be removed if Q/gate are separated. ggml_cont is not recommended when dealing with big tensors

Actually none of these need ggml_cont, Q is 3D already, Q/K are RoPEd so can be views and V can also be a 3D view now.

Edit: sorry, not quite true about V, only if QKV is fused, the weird gate fuse threw me off. Nevertheless, K/V are already contiguous at this point.

the problem is that Q is non-contiguous and ggml_rope(_ext) does not work very well with non-cont tensors, it's still buggy on certain backends

the problem is that Q is non-contiguous and ggml_rope(_ext) does not work very well with non-cont tensors, it's still buggy on certain backends

Are you sure? AFAIK those issues are fixed.

Edit: Also, if there still are issues they will never get fixed if we work around them. :)

the problem is that Q is non-contiguous and ggml_rope(_ext) does not work very well with non-cont tensors, it's still buggy on certain backends

I think all of these cases are fixed now.

This was an impl of 2D rope that relies on ggml_view: https://github.com/ngxson/ggml-easy/blob/f56e5e499b1f21a4aae73010e9d9582840428457/demo/2d-rope.cpp

It works on CPU and Metal, but doesn't work on CUDA/Vulkan. Couldn't tested on other backends, but feel free to make a PR to address this issue.

Does it still fail? I think these PRs should have addressed the problem:

cuda : fix rope with partial rotation and non-cont src #14580

vulkan: fix rope with partial rotation and non-cont src #14582

yes that seems to work. sorry @pwilkin you will need to manually revert the change where I split Q/gate. the tensor shape for Q will be:

layer.wq = create_tensor(tn(LLM_TENSOR_ATTN_Q, "weight", i), { n_embd, n_embd_head_k * n_head * 2 }, 0);

ngxson · 2025-09-20T05:02:45Z

src/llama-model.cpp

+                            layer.ssm_a = create_tensor(tn(LLM_TENSOR_SSM_A, i), { hparams.ssm_dt_rank }, 0);
+                            layer.ssm_beta_alpha = create_tensor(tn(LLM_TENSOR_SSM_BETA_ALPHA, "weight", i), { n_embd, ba_projection_size }, 0);
+                            layer.ssm_norm = create_tensor(tn(LLM_TENSOR_SSM_NORM, "weight", i), { head_v_dim }, 0);
+                            layer.ssm_out = create_tensor(tn(LLM_TENSOR_SSM_OUT, "weight", i), { n_ff, n_embd }, 0);


Shape of LLM_TENSOR_ATTN_Q and LLM_TENSOR_SSM_OUT should not contain n_ff

ngxson · 2025-09-20T05:03:28Z

^ proposed fix for the 3 comments above: 46110e0

src/llama-model.cpp

ggml/src/ggml-cpu/ggml-cpu.c

pwilkin · 2025-09-20T10:26:32Z

@ngxson Thanks, scale_bias was one op I was missing in my endeavors :>

I got an LLM to rewrite the internal delta into tensor logic. After a day of manually fixing that crap, I think I understand it enough to rewrite it myself ;)

ngxson · 2025-09-20T10:31:46Z

Honestly I would prefer taking time to understand the mamba/ssm implementation then writing the code manually. Code written by LLM are mostly attempts for 1-to-1 translation from pytorch --> GGML which looks quite confusing

pwilkin · 2025-09-20T12:14:57Z

Honestly I would prefer taking time to understand the mamba/ssm implementation then writing the code manually. Code written by LLM are mostly attempts for 1-to-1 translation from pytorch --> GGML which looks quite confusing

Yeah, for me getting a rough outline then going over it manually is the best way to learn :)

I tried the "one-to-one" approach and ended up with a graph that wouldn't fit in 16 GB of RAM for a 500M model...

pwilkin · 2025-09-20T21:22:09Z

Aight, I cleaned up the main graph calculation, now I have to figure out how to include conv_states_all in my delta_net function in order to not get the memory error.

…capture.

pwilkin · 2025-11-26T16:54:12Z

@ggerganov I readded <cmath>, as it's used by quite a lot of models that import models.h, since clang standard doesn't like transitive includes we'll need to move that to the respective model files, but I think that's better done in a separate PR.

Tested perplexity, is fine on both CUDA and --cpu-moe, tested generation (asked him to write me a crazy Haskell game inspired by quants.c), everything seems to be fine.

EDIT: bonus - quantum cats: https://gist.github.com/pwilkin/108904f9084250b0329f9dc419ff019e

convert_hf_to_gguf.py

CISC · 2025-11-26T20:44:07Z

src/models/qwen3next.cpp

+        ggml_tensor * moe_out =
+            build_moe_ffn(cur, model.layers[il].ffn_gate_inp, model.layers[il].ffn_up_exps,
+                          model.layers[il].ffn_gate_exps, model.layers[il].ffn_down_exps, nullptr, n_expert,
+                          n_expert_used, LLM_FFN_SILU, true, false, 0.0, LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX, il);


Forgot this one. :)

Co-authored-by: Sigbjørn Skjæret <[email protected]>

convert_hf_to_gguf.py

lovedheart · 2025-11-27T13:26:53Z

src/llama-model.cpp

+                for (uint32_t i = 0; i < hparams.n_layer; ++i) {
+                    hparams.recurrent_layer_arr[i] = ((i + 1) % 4 != 0); // TODO: extract the magic 4 from "full_attention_interval"
+                }
+


I think the n_layer should be 48 here.

CISC · 2025-11-28T09:06:42Z

I think today is the big merge day, unless someone has anything to add, let's merge in an hour.

pwilkin · 2025-11-28T10:52:52Z

@CISC do it! Press the button! Let it explode!

CISC · 2025-11-28T11:03:22Z

💣 💥

nvdg · 2025-11-28T11:06:49Z

Thanks for all the hard work everyone! 💪

gopinath87607 · 2025-11-28T11:07:23Z

big thanks to everyone !!!

pwilkin force-pushed the qwen3_next branch from 2f47798 to 178230e Compare September 18, 2025 19:47

github-actions bot added python python script changes ggml changes relating to the ggml tensor library for machine learning labels Sep 18, 2025

ngxson reviewed Sep 19, 2025

View reviewed changes

pwilkin marked this pull request as draft September 19, 2025 08:07

github-actions bot added the examples label Sep 19, 2025

rick-github mentioned this pull request Sep 19, 2025

Qwen3-Next ollama/ollama#12349

Open

ngxson reviewed Sep 20, 2025

View reviewed changes

src/llama-model.cpp Outdated Show resolved Hide resolved

ngxson reviewed Sep 20, 2025

View reviewed changes

src/llama-model.cpp Outdated Show resolved Hide resolved

ngxson reviewed Sep 20, 2025

View reviewed changes

src/llama-model.cpp Outdated Show resolved Hide resolved

ngxson reviewed Sep 20, 2025

View reviewed changes

ggml/src/ggml-cpu/ggml-cpu.c Outdated Show resolved Hide resolved

ngxson mentioned this pull request Sep 20, 2025

mtmd: more optimized build_rope_2d #16126

Closed

pwilkin and others added 11 commits November 26, 2025 17:40

Nail in the coffin?

b4a62f7

Fix Windows build

9815b0c

Eh, one fails on Windows, the other fails on Mac... just use general …

6bd04d9

…capture.

quant : cleanup

6bb8c34

model : cleanup

32f8e40

qwen3 : cleanup

5e0e261

cont : cleanup

d160879

cont : cleanup

6766ac0

ggml : revert change

6c26596

qwen3 : cleanup

2b36efa

cont : cleanup

c4cf07b

pwilkin force-pushed the qwen3_next branch from e43a910 to c4cf07b Compare November 26, 2025 16:40

Readd cmath

5c3a0d7

CISC approved these changes Nov 26, 2025

View reviewed changes

qwen3 : fix typo

0dda56d

github-actions bot mentioned this pull request Nov 27, 2025

Reddit News Daily 2025-11-27 gitlawr/reddit-daily-news#76

Open

pwilkin and others added 2 commits November 27, 2025 08:08

Update convert_hf_to_gguf.py

d6b3150

Co-authored-by: Sigbjørn Skjæret <[email protected]>

Usual suspects

620d351

CISC reviewed Nov 27, 2025

View reviewed changes

convert_hf_to_gguf.py Outdated Show resolved Hide resolved

fix my bad suggestion

370b42a

lovedheart reviewed Nov 27, 2025

View reviewed changes

CISC merged commit ff55414 into ggml-org:master Nov 28, 2025
74 of 78 checks passed

Kenshiro-28 mentioned this pull request Nov 28, 2025

Add support for Qwen Next abetlen/llama-cpp-python#2097

Open

zeerd mentioned this pull request Nov 28, 2025

Support for qwen3next architecture lmstudio-ai/lmstudio-bug-tracker#1214

Open

Model: Qwen3 Next #16095

Model: Qwen3 Next #16095

Conversation

pwilkin commented Sep 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gabe-l-hart commented Sep 18, 2025

Uh oh!

bugparty commented Sep 18, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

pwilkin commented Sep 19, 2025

Uh oh!

CISC commented Sep 19, 2025

Uh oh!

ngxson commented Sep 19, 2025

Uh oh!

pwilkin commented Sep 19, 2025

Uh oh!

pwilkin commented Sep 19, 2025

Uh oh!

yarikdevcom commented Sep 19, 2025

Uh oh!

pwilkin commented Sep 19, 2025

Uh oh!

ServeurpersoCom commented Sep 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ngxson commented Sep 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

CISC Sep 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

CISC Sep 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ngxson commented Sep 20, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

pwilkin commented Sep 20, 2025

Uh oh!

ngxson commented Sep 20, 2025

Uh oh!

pwilkin commented Sep 20, 2025

Uh oh!

pwilkin commented Sep 20, 2025

Uh oh!

pwilkin commented Nov 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

pwilkin commented Sep 18, 2025 •

edited

Loading

ServeurpersoCom commented Sep 19, 2025 •

edited

Loading

ngxson commented Sep 20, 2025 •

edited

Loading

CISC Sep 20, 2025 •

edited

Loading

CISC Sep 20, 2025 •

edited

Loading

pwilkin commented Nov 26, 2025 •

edited

Loading