biondizzle

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-10 15:01:30 +00:00

c1b013234e Fix cache-bust: embed VLLM_COMMIT in git clone RUN command

biondizzle pushed to cmm at biondizzle/vllm

2026-04-10 14:58:59 +00:00

487dd34e04 Selective prefetch: only prefetch allocations <2 GiB to GPU

biondizzle pushed to master at biondizzle/smollm3-3b-vllm

2026-04-10 14:05:53 +00:00

57870d5a51 override parser

biondizzle pushed to master at biondizzle/smollm3-3b-vllm

2026-04-10 14:02:05 +00:00

a5bb29930f add in lora

biondizzle created branch master in biondizzle/smollm3-3b-vllm

2026-04-10 13:56:02 +00:00

biondizzle pushed to master at biondizzle/smollm3-3b-vllm

2026-04-10 13:56:02 +00:00

5029ab3b40 init commit

biondizzle created repository biondizzle/smollm3-3b-vllm

2026-04-10 13:34:05 +00:00

biondizzle pushed to master at biondizzle/smollora

2026-04-10 06:37:44 +00:00

eb0850bca6 Fix: correct source dataset tag names for tool_call/tool_response

biondizzle pushed to master at biondizzle/smollora

2026-04-10 06:33:11 +00:00

99481ca127 Add HF_TOKEN env var

biondizzle pushed to master at biondizzle/smollora

2026-04-10 06:32:56 +00:00

9e723d393e Fix: remove unsupported --model arg from prepare_data.py call

biondizzle pushed to master at biondizzle/smollora

2026-04-10 06:24:51 +00:00

adbd85366b init commit

46a3ddbb25 Add deployment runbook

6af62c85d5 Add docker-compose with /srv mounts for persistent data

82348341b0 Initial LoRA training setup for SmolLM3-3B tool calling

biondizzle created branch master in biondizzle/smollora

2026-04-10 06:24:50 +00:00

biondizzle created repository biondizzle/smollora

2026-04-10 06:22:04 +00:00

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-10 06:01:23 +00:00

98b4ae6676 Add VLLM_COMMIT cache-bust arg to Dockerfile

biondizzle pushed to cmm at biondizzle/vllm

2026-04-10 05:58:22 +00:00

a15f86ecfa Remove cudaMemPrefetchAsync from managed allocator

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-10 02:45:07 +00:00

c583bcb4fc Fix cudaMemPrefetchAsync for CUDA 13: use cudaMemLocation + flags=0 (no stream param)

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-10 01:48:03 +00:00

6053e6d0ea Fix cudaMemPrefetchAsync: use int device instead of cudaMemLocation struct

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-09 23:25:51 +00:00

aadde3ddf9 CMM: Fix OOM and subprocess crashes for GH200 EGM

biondizzle pushed to cuda-malloc-managed at biondizzle/grace-gpu-containers

2026-04-09 22:05:42 +00:00

079eb88d7d Switch vLLM source to Gitea fork (cmm branch)

biondizzle pushed tag v0.9.1 to biondizzle/vllm

2026-04-09 22:04:42 +00:00