GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
samarthraina
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
samarthraina
Model Tree
teknium/OpenHermes-2.5-Mistral-7B
Input Modalities
Output Modalities
Supported Functionality
teknium/OpenHermes-2.5-Mistral-7B, revision 24c0bea14d53e6f67f1fbe2eca5bfe7cae389b33mistralai/Mistral-7B-v0.1harmless-baseq_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_projembed_tokens / lm_head in
float16. The base model itself is bfloat16, which is what the snippet below loads.Full hyperparameters are in training_args.json.
Requires
transformers >= 5.0. The tokenizer in this repository is stored in the currenttokenizers-backend format. Loading it withtransformers4.x raisesTokenizer class TokenizersBackend does not exist.
The training tokenizer in this repository has one additional token relative to the base
model — a dedicated <|pad|> padding token:
| tokens | |
|---|---|
base teknium/OpenHermes-2.5-Mistral-7B embedding table | 32,002 |
| adapter tokenizer in this repository | 32,003 |
The base embedding table must be resized to match before the adapter is attached, or loading fails on a shape mismatch. The snippet below does this correctly.
This repository ships the trained embed_tokens and lm_head weights at the resized
[32003, 4096] shape alongside the LoRA tensors, so once you resize, the added token's
values come from the adapter rather than from random initialisation.
python
import torchfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModelADAPTER = "samarthraina/D-STEER-Mistral-7B-OpenHermes-DPO-Adapter"BASE = "teknium/OpenHermes-2.5-Mistral-7B"BASE_REVISION = "24c0bea14d53e6f67f1fbe2eca5bfe7cae389b33"# 1. load the adapter's tokenizer (it carries the extra token)tokenizer = AutoTokenizer.from_pretrained(ADAPTER)# 2. load the base model at the pinned revisionmodel = AutoModelForCausalLM.from_pretrained(BASE, revision=BASE_REVISION, dtype=torch.bfloat16)# 3. check the embedding table against the tokenizern_embed = model.get_input_embeddings().weight.shape[0]print(f"base embeddings: {n_embed} | tokenizer: {len(tokenizer)}")# 4. resize only if they disagreeif n_embed != len(tokenizer):model.resize_token_embeddings(len(tokenizer))# 5. attach the LoRA adaptermodel = PeftModel.from_pretrained(model, ADAPTER)model.eval()# 6. optional: fold the adapter into the base weights# model = model.merge_and_unload()
Released for AI-safety research: as the preference-tuned endpoint of a matched pair for estimating and evaluating harmlessness directions in activation space.
| Item | Value |
|---|---|
adapter_model.safetensors SHA-256 | 79c81ebc54c040c305fd35524fafe0edb3facc6be91284ad3e5ddc97bb758517 |
adapter_model.safetensors size | 692,169,616 bytes |
| D-STEER code commit | 84b50df41dd83fee78de149e6f849cb8e09c8c8e |
| Model artifact manifest SHA-256 | a16d842cb1fc0212a0a081b4c49183b2053e5902b00b90dfda255992a1a6a609 |
training_args.bin SHA-256 (recorded, not published) | see |
Apache-2.0, inherited from the base model
teknium/OpenHermes-2.5-Mistral-7B (Apache-2.0), which is itself
derived from mistralai/Mistral-7B-v0.1
(Apache-2.0). This adapter contains no Llama-derived weights.
PROVENANCE.jsonPROVENANCE.json in this repository records the SHA-256 of every published file. |