为了安全，强烈建议开启2FA双因子认证：User Settings -> Account -> Enable two-factor authentication！！！

Tags

Tags give the ability to mark specific points in history as being important

b5329

611aa914 · metal : optimize MoE for large batches (#13388) · May 09, 2025
b5328

0cf6725e · CUDA: FA support for Deepseek (Ampere or newer) (#13306) · May 09, 2025
b5327

27ebfcac · llama : do not crash if there is no CPU backend (#13395) · May 09, 2025
b5326

5c86c9ed · CUDA: fix crash on large batch size for MoE models (#13384) · May 09, 2025
b5325

efb8b47e · imatrix : Add --parse-special for enabling parsing of special tokens in... · May 09, 2025
b5324

0527771d · llama-run: add support for downloading models from ModelScope (#13370) · May 09, 2025
b5323

2189fd3b · mtmd : fix batch_view for m-rope (#13397) · May 09, 2025
b5322

3f96aeff · llama : one-off chat template fix for Mistral-Small-2503 (#13398) · May 09, 2025
b5321

b486ba05 · rpc : add rpc_msg_set_tensor_hash_req (#13353) · May 09, 2025
b5320

02115dcd · vulkan: Allow up to 4096 elements for mul_mat_id row_ids (#13326) · May 09, 2025
b5318

15e03282 · ci : limit write permission to only the release step + fixes (#13392) · May 08, 2025
b5317

f05a6d71 · mtmd : Expose helper_decode_image_chunk (#13366) · May 08, 2025
b5315

8c83449c · server : (webui) revamp the input area, plus many small UI improvements (#13365) · May 08, 2025
b5313

0ccc1213 · mtmd : fix the calculation of n_tokens for smolvlm (#13381) · May 08, 2025
b5311

51fb96b1 · context : remove logits_all flag (#13284) · May 08, 2025
b5310

70a6991e · ci : move release workflow to a separate file (#13362) · May 08, 2025
b5309

f0610212 · llama : print size and type of overridden tensors (#13364) · May 08, 2025
b5308

8733e0cf · sycl: addressing non-contiguous src1 mul_mats (nc and batched) (#13343) · May 08, 2025
b5306

d8794338 · sync : ggml · May 07, 2025
b5303

bc4e1128 · llama : deci : support ffn-free with attention (#13296) · May 07, 2025

Prev
1
2
3
4
5
6
…
178
Next