24 Commits

Author SHA1 Message Date
Alex Cheema
3c7bd48aa3 get rid of some more hf bloat 2025-01-27 01:08:46 +00:00
Alex Cheema
a3766f538a add exception for mlx-community/DeepSeek-R1-3bit and mlx-community/DeepSeek-V3-3bit in tokenizers test 2025-01-26 20:32:48 +00:00
Alex Cheema
fdd05baddb fix tokenizer tests 2025-01-24 18:13:36 +00:00
Alex Cheema
e7b98f5ae5 fix unit tests 2025-01-12 03:35:24 +00:00
Alex Cheema
40696b21f7 typo in phi test 2025-01-08 22:09:04 +00:00
Alex Cheema
2846a9122f tok tests 2025-01-08 16:39:11 +00:00
DeftDawg
cde912deef - Use #!/usr/bin/env bash instead of #!/bin/bash for better portability 2024-12-22 01:14:54 -05:00
Alex Cheema
b0e079b36a fix counts in testmodelhelpers 2024-12-13 17:46:44 +00:00
Alex Cheema
01e6b931a0 fix modelpool, add tests in test/test_model_helpers.py 2024-11-18 20:52:06 +04:00
Alex Cheema
f2d5beee08 change chatgpt api port from 8000 to 52415 2024-11-16 10:58:55 +04:00
Alex Cheema
34f3c4a155 fix tokenizers test with restructured models 2024-11-13 17:06:11 +04:00
Alex Cheema
472359147d ignore 8bit llama 405b from tokenizers test 2024-11-10 13:54:10 +04:00
Alex Cheema
e80ed76c03 ignore in test-tokenizers 2024-11-06 22:16:13 +04:00
Alex Cheema
bc1d88d86d ignore dummy 2024-10-25 15:15:37 -07:00
Alex Cheema
b6d239af49 ignore deepseek v2.5 from tokenizers test as it requires remote code 2024-09-20 18:38:07 +01:00
Alex Cheema
68028cc980 ignore Qwen models in tokenizers test until bos issue is fixed 2024-09-19 00:18:21 +01:00
Alex Cheema
84187113de add a test for hf get_weight_map 2024-09-05 16:39:40 +01:00
Alex Cheema
f342cdcae9 get rid of -secs suffix 2024-09-04 16:49:23 +01:00
Alex Cheema
355c579965 more robust discovery / peer handling. now we track if the same node id changes address, then we immediately conenct to it 2024-09-04 15:29:29 +01:00
Alex Cheema
dcb3ac76a8 test kill pids 2024-09-04 14:21:29 +01:00
Alex Cheema
15b5043d6e test for reconnect 2024-09-04 13:11:53 +01:00
Alex Cheema
1f9d16ec78 run tokenizers test in ci, run all models available 2024-08-23 16:35:33 +01:00
Alex Cheema
710e5a31e7 TODO for why use_fast=False is giving inconsistent behaviour (no spaces decoding invididual tokens) for Mistral-Large-Instruct-2407-4bit 2024-08-22 14:45:08 +01:00
Alex Cheema
e17e5f9a41 tests for tokenizers. unfortunately use_fast=False and use_fast=True give different behaviour 2024-08-22 14:44:59 +01:00